Let us now think about the other extreme of the market: luxury goods that are built in one place and shipped around the world, such as luxury cars. All the Ferraris sold around the world are assembled in Italy. And they are so expensive that pure local distribution costs are a relatively small fraction of the ultimate price tag. On the other hand, Ferrari prices are impacted by transportation costs, taxes, and marketing decisions.
The MSRP of a baseline Ferrari 488 GTB in the US is $252,800. The same car costs about €226,039 in France, which in August 2018 was about $262,380, but some taxes are included in France and not in the US. Broadly speaking, then, we can say that the prices are similar, even though the US has a higher per-capita income than France. The 488 GTB also costs about the same in Prague as in Paris, even though per-capita GDP in the Czech Republic is 40 percent lower than in France. This is consistent with basic economic theory. Ferraris are traded goods, and their cost does not depend much on where they are sold. So we can expect to see about the same prices in Prague and in New York City.
Perhaps surprisingly, however, the same car would cost more than twice as much in China, about $550,000. Why? China imposes import duties and complicated taxes. For instance, China imposes higher taxes on cars with an engine capacity of more than four liters (engines mostly found in high-end luxury cars … and US gas guzzlers). Probably not by coincidence, the engine of the new 488 GTB is 3.9 liters, just below the four-liter threshold. One has to take these details into account to obtain an accurate comparison of prices. Even after we adjust for taxes, however, prices of Ferraris and other high-end luxury cars are much higher in China than in the US or Europe. The common explanation is that Chinese buyers are more willing to spend large amounts of money on exclusive cars than buyers in Europe or in the US because the effect of owning a status symbol is stronger. In our economic framework, we would say that wealthy Chinese buyers have a high willingness to pay and a low elasticity of substitution. Therefore, as economic theory would predict, markups are high. That’s Econ 101, exactly as in Chapter 1, but what is surprising is the sheer magnitude of the difference.
Let us keep these two examples in mind when we compare prices around the world. It is quite clear that the price of a haircut has a lot to do with average local wages, and that the price of a Ferrari has almost nothing to do with local wages or GDP per capita. Most goods and services fall in between these two extreme examples: they are part haircuts and part Ferraris.
Are You Big Mac Richer, PPP Richer, or Market Rate Richer?
Prices matter a lot in economics. In a sense, economics is really the science of figuring out equilibrium prices. In general, the higher the price of something, the less of it one would want to buy. That’s what we call a demand curve. But that assumes a degree of consumer choice, meaning that the buyer is free to choose to consume less, or to substitute a different good for the one in question. If a consumer does not have access to alternative products and cannot meaningfully reduce her consumption—say on basic staples like food and shelter—then higher prices create higher poverty. In general, higher prices have two effects: they lead to lower demand and to lower standards of living.
As it turns out, however, even seemingly concrete concepts like “getting richer” or “getting poorer” have very different meanings, depending on context.
To compare incomes around the world, we need to convert them from local currencies into a common currency, using either foreign exchange market rates (FOREX) or purchasing power parity exchange rates (PPP). The FOREX exchange rate is the one that you read about in the newspapers or online. It is the rate that says, for instance, that 1 euro equals 1.2 dollars. The PPP exchange rates are more complicated and more interesting. PPP rates are defined in such a way that one unit of currency can purchase the same amount of goods and services everywhere.
Let us illustrate the idea with an example. Pierre is from Bordeaux and earns €50,000 per year. Karen is from Boston and earns $70,000 per year. Who is richer? It depends on their consumption baskets. Suppose they both spend one year living in the Azores. Pierre’s income in the Azores is €50,000. Karen’s $70k is worth 70k / 1.2 = €58,333. In the Azores, Karen is clearly richer.
We can compare GDP per capital around the world by converting all incomes into a common currency using FOREX rates. For instance, US GDP per capita in 2017 was about $53,130. France GDP per capita in 2017 was €35,400. Suppose the euro trades at $1.2. France’s GDP per capita at market rate was therefore $42,480. In that sense, French people are only 80 percent as rich as American people. That is a good measure of how they would feel as tourists in the same foreign country.
But is that the right way to think about their relative standards of living? It only makes sense to compare people’s income at market exchange rates if they live in the same place. If they live in different places instead, we need to consider the fact that they have different needs (think of heating costs in Anchorage versus San Juan) and that they face different local prices. If Pierre lives in France, he pays $35 per month for his broadband internet connection, and Karen pays $80 per month in the US. With his income, Pierre can buy more internet access than Karen. In that sense, we would say Pierre is richer.
A more useful way to assess who is richer, then, would be to compute income at purchasing power parity (PPP) rates, i.e., income divided by the price of a common set of goods and services. This is easier said than done, as you can imagine. For a start, Pierre and Karen consume many goods and services, not just goods and services accessed on the internet. When we make international comparisons of real income, we need to look at the prices of the items that people actually consume. So how do we proceed?
The first thing we can do is to focus on items that everyone consumes—or at least, that are sold everywhere. In 1986, the Economist invented the Big Mac index. It was half a joke and half an attempt to make the theory of PPP more digestible (pun intended). They went around the world and collected prices of the same McDonald’s sandwich.
This may seem frivolous, but it is actually quite useful, because McDonald’s requires a large degree of consistency in its products, regardless of where they are made and sold. That means all the inputs that go into a Big Mac are virtually identical, whether it’s made in Paris or Paducah.
TABLE 7.1
FOREX Rates, Big Mac Prices, and ICP PPP Rates
Year
Market exchange rate
Local price of Big Mac
PPP exchange rates, €1 = $x
€1 = $x
EA19
US
Big Mac
ICP
2000
$0.92
€2.56
$2.51
$0.98
$1.16
2001
$0.89
€2.57
$2.54
$0.99
$1.16
2002
$0.94
€2.67
$2.49
$0.93
$1.17
2003
$1.13
€2.71
$2.71
$1.00
$1.16
2004
$1.24
€2.74
$2.90
$1.06
$1.17
2005
$1.24
€2.92
$3.06
$1.05
$1.17
2006
$1.25
€2.93
$3.15
$1.08
$1.21
2007
$1.37
€3.06
$3.41
$1.11
$1.22
2008
$1.46
€3.37
$3.57
$1.06
$1.24
2009
$1.39
€3.31
$3.57
$1.08
$1.26
2010
$1.32
<
br /> €3.38
$3.73
$1.10
$1.26
2011
$1.39
€3.44
$4.06
$1.18
$1.28
2012
$1.28
€3.58
$4.33
$1.21
$1.29
2013
$1.33
€3.62
$4.56
$1.26
$1.32
2014
$1.33
€3.68
$4.79
$1.30
$1.33
2015
$1.11
€3.70
$4.79
$1.29
$1.32
2016
$1.11
€3.82
$5.04
$1.32
$1.33
2017
$1.13
€3.91
$5.30
$1.36
$1.33
Source: Economist, OECD
Table 7.1 shows you the Big Mac prices for the US and the euro area. In 2003, a Big Mac cost €2.71 (on average) in the EA19 and $2.71 in the US. Clearly, in terms of Big Macs, the euro and the dollar had the same purchasing power. We would say that the purchasing power parity (PPP) Big Mac exchange rate was $1 per euro. However, in that same year, the currency markets valued the euro at $1.13. In that sense, the euro as a financial asset looked a bit expensive, or “overvalued” as we would say in the language of international economics.
You might reasonably object that the Big Mac index is too narrow. People (thankfully) do not consume only Big Macs. Economists have built other indexes that attempt to do something similar on a larger scale.
The United Nations, together with the University of Pennsylvania, established the International Comparisons Program (ICP) in 1968. The goal of the ICP is to facilitate price comparisons across countries.a They conduct global surveys of prices (in about 150 countries and for 1,000 products) to compute their estimate of PPP exchange rates. In 2003, the ICP PPP was $1.16 per euro. Using this measure, the financial value of the euro ($1.13 in 2003) was about right, even a bit cheap.
In 2007, ICP PPP was $1.22 per euro. Big Mac PPP was only $1.11, and the nominal exchange rate was $1.37. Clearly, these three indexes evolve somewhat independently. The correlations between these three exchange rates reveal an interesting pattern, however.
The two PPP measures—Big Mac and ICP—are both trending up, meaning that, in general, the same amount of money buys more in Europe than it does in the US. The trend is steeper and bumpier for the Big Mac PPP, as one would expect since it is based on only one local good. Still, I would argue that both PPP rates tell roughly the same story.
On the other hand, the market exchange rate has a weak correlation with either PPP exchange rate of only about 0.38. Clearly, the forces that pin down the euro / dollar rate on the FOREX market are different from those that influence the price of the typical consumer’s basket.b FOREX prices are much more influenced by financial conditions, interest rates, risk appetites, and so on. Given our long-term perspective and our focus on the real economy, PPP exchange rates are clearly more relevant for our analysis.
Prices, Marginal Costs, and Markups
Figure 7.1 shows that prices have been going up faster in the US than in Europe over the past eighteen years. ICP PPP went from $1.16 to $1.33, so prices went up by 15 percent more in the US than in Europe over this period. Why?
As we have discussed, we can think of the price of a good as a markup over its production cost:
Price = (1 + markup) × MC
MC is the marginal cost—the cost of the last unit of production. The marginal cost depends on the cost of labor (higher wages = higher costs) and on productivity (higher productivity = lower costs). The price can also depend on taxes and other input costs beyond labor, such as energy and raw materials. Can higher prices in the US be explained by a combination of the following forces?
Wages have increased more in the US;
Markups have increased more in the US;
Productivity growth has been faster in Europe;
Taxes or energy costs have decreased more in Europe.
FIGURE 7.1 Nominal euro / dollar exchange rates
I am going to ignore the last explanation for its implausibility. Taxes have certainly not decreased in Europe relative to the US over this period. With respect to energy, the US has benefited from the boom in shale gas extraction, so its internal energy costs must have decreased relative to Europe. And since the nominal exchange rate is the same today as in 2012, oil import prices cannot explain the difference.
Productivity raises a more complex set of issues. In theory, it is clearly better to control for productivity differences. But productivity is hard to measure and requires a host of auxiliary assumptions, so there is the risk of adding noise or even biases if the measurement errors in productivity and prices are not independent.
Fortunately, it turns out that the results hold with or without controlling for productivity. To some extent, this reflects the fact that measures of GDP per capita follow rather similar trends. If anything, the US has done little better than the euro area over the past twenty years in terms of productivity growth. Controlling for productivity therefore only reinforces my point.
We can construct a measure of marginal costs with wages controlling for productivity. This is what economists call unit labor costs (ULC), as defined in Box 7.1. Let us compare the evolution of markups in the US and in Europe. Box 7.1 explains how to construct a measure of the change in the markup between year 2000 and any year t afterward for any country i: DMi,t. Now we want to compare the relative evolution of markups between Europe and the US, so we simply compute the difference:
RDMi,t = DMi,t − DMUS,t
The solid gray line in Figure 7.2 shows the weighted average of RDMi,t across the ten main EU countries for which we can also compute measures of concentration. The line with circles shows the average change in the CR4 (remember, that’s the market share of the four largest firms in an industry) in the same EU countries, relative to the US. Over the same period, CR4 has increased by 5 percentage points in the US relative to the EU. In practice, as we saw earlier, concentration has remained constant in Europe but has increased by about 5 points in the US.
Box 7.1. Unit Labor Costs and Markups
Unit labor costs (ULC) measure the average cost of labor per unit of output. We compute them as the ratio of total labor costs to real output. In other words, the unit labor cost is:
ULC = WL / Y
In that equation, Y is real output and WL measures total labor costs. By definition, WL is the product of the average wage (W) times the amount of labor employed (L), which can be proxied either by the number of employees or by the total number of hours worked.
Equivalently, we can think of the unit labor cost as the wage relative to the productivity of labor. We define labor productivity (LP) as output per worker (or per hour):
LP = Y / L
Using this alternative definition, the unit labor cost is also the cost of one unit of labor (the wage) divided by the productivity of that unit of labor. The unit labor cost in country i at time t is then:
ULCi,t = Wi,t / LPi,t
We can define the logarithm of the markup of prices over unit labor costs as
Mi,t = log(Pi,t) − log(ULCi,t)
Differences in the accounting between countries can create persistent differences between measured prices and wages. We can remove these differences by considering the difference from a base year, which we call time 0, set here to the year 2000:
DMi,t = Mi,t − Mi,0
DMi,t then gives us the evolution of markups in country i over time.
You can see that markups have decreased by about 14 percent in Europe relative to the US. Remember that the total increase in relative US prices was 15 percent, and the increase in relative wages
about 7 percent, so the price / wage markup has increased by about 8 percent more in the US than in Europe. How do we get a 14 percent markup increase? In the industry data that we use for the benchmark, productivity rises by 6 percent more in the US than in Europe. In theory, higher productivity should have led to either lower prices or higher wages in the US. It did not, so that implies an even higher markup of 8 percent + 6 percent = 14 percent. As I have explained above, measuring productivity is tricky, and we need to treat this 6 percent number with a (large) grain of salt. But even if we ignore it entirely, we would still get a relative increase in US markups of about 8 percent.
FIGURE 7.2 Markup and concentration in Europe versus the US
Figure 7.2 is what we call time series evidence: we consider the evolution of two series of data over time. We see concentration going down in Europe relative to the US, and then we see European markups decrease relative to American markups.
Is this a smoking gun? Yes and no. It is pretty convincing because this is direct evidence that prices have increased more in the US than in the EU for the same goods and services. On the other hand, the correlation between markups and concentration might be a coincidence. It’s not likely, but it is possible. In a sense, the time series contain only one observation. To have a smoking gun we would have to be sure that nothing was contributing to the disparity in the price / wage markup numbers and concentration at the same time. Given the complexities of these two economies, it is difficult to effectively exclude all other potential explanations.
To gain confidence in our diagnosis, we would like to see this pattern repeated in different samples. We can look inside Europe and ask if we see the same pattern across countries and time periods. Is it true that the change in concentration predicts the change in markups across years and across countries? To run this test, we have what we call a panel data set. It contains data for ten countries over sixteen years. This means we can compare Germany and Italy in a given year, or we can compare Germany to itself a few years earlier. If our theory is correct, we would expect a positive relationship between concentration and markups across time and space. In our work, Gutiérrez and I (2018a) find that this is indeed the case.
The Great Reversal Page 13