Europe's Productivity Debate Needs Better Data

Krugman is right that Europe's relative living standards have held up better than the decline narrative suggests. His critics are right that America has built a lead in intangible capital and tech.

The recent exchange between Paul Krugman, Luis Garicano, Philippe Aghion and Antonin Bergeaud has turned Europe’s productivity gap with the United States into a dispute over what the data actually measure. Using industry-level growth accounting data, I take a slightly different approach.

Europe Versus America: A Response to the Critics

At the risk of oversimplifying, the two positions can be summarised as follows.

Krugman’s argument is that much of the apparent divergence disappears once output is valued at current prices and converted using purchasing power parities. If one economy becomes much better at producing a particular class of goods, their relative price tends to fall. The gains from that increased productive capacity are then shared much more broadly than the location of production itself. American advances in software, digital services and technology raise living standards well beyond the United States. On measures that focus on what people can actually consume, Europe looks considerably closer to America than many decline narratives imply.

Garicano’s objection is that this risks throwing away information that matters. Constant price measures are designed precisely to capture changes in productive capacity over time. They show that the United States has expanded output in technology-intensive sectors far more rapidly than Europe, regardless of what has happened to relative prices. An economy that develops new technologies, business models and productive capabilities has become more capable in a meaningful sense, even if the resulting products become cheaper.

In practice, the two arguments are less contradictory than they first appear. Krugman is largely talking about relative living standards. Garicano is talking about productive capacity and technological dynamism. Those are related, but they are not the same thing. An economy can maintain relatively high living standards while gradually losing ground in the sectors that drive future productivity growth.

The question, then, is not whether one side is right and the other wrong. It is whether the data can tell us something about the source of the divergence. Using industry-level growth accounting data, the answer appears to be yes. Krugman is broadly correct that Europe’s relative living standards have held up better than many headline productivity comparisons suggest. But the residual gap that remains reflects a genuine divergence in the accumulation of intangible capital and in the sectors most closely associated with technological progress. In that sense, both sides are identifying something real. They are simply looking at different parts of a similar story.

A different way of measuring the gap

Most contributions to this debate rely on aggregate GDP or productivity measures. Those are useful for establishing that a gap exists, but they tell us relatively little about where it comes from. They cannot tell us which sectors are driving the divergence, whether it reflects differences in technology, capital accumulation or labour inputs, or how much of it is concentrated in a handful of industries. To answer those questions, you need to look beneath the aggregates.

The data used here come from EUKLEMS & INTANProd, a growth accounting database covering EU member states, the United States, the United Kingdom and Japan. Unlike standard productivity datasets, it decomposes growth into its underlying sources across roughly sixty industries. For each sector, it reports output, prices, labour inputs, capital inputs and total factor productivity.

That makes it possible to evaluate one of Krugman’s central claims directly. If the United States has pulled ahead primarily in sectors characterised by rapid technological progress and falling relative prices, the data should show it. The gap should be concentrated in a relatively narrow set of industries rather than spread evenly across the economy.

The second reason this dataset is useful is that it incorporates detailed measures of intangible capital. These include not only research and development and software, but also organisational capital, design, branding and firm-specific human capital. That matters because much of the disagreement ultimately revolves around whether the United States has developed productive capabilities that conventional measures fail to capture adequately.

Technology, prices and the growth gap

The easiest way to see what Krugman is talking about is to look at the tech sector, defined here as computer and electronics manufacturing (C26) together with information and communication services (J), and compare the volume of output it produces and its share of the economy measured at current prices.

Real output almost quadruples over the period. Yet the sector’s share of nominal GDP barely changes, rising only marginally from around 8% to 8.5%. Real production increases while the sector's share of nominal GDP remains almost unchanged. This lends support to Krugman’s claim that gains are largely passed on through falling prices rather than captured as a larger share of the economy's market value.

Technological progress increases productive capacity, but it also makes many of the resulting goods and services cheaper. A measure based on constant prices records a large increase in output and productivity. A measure based on current prices records a much smaller change because the gains are partly passed on through lower prices.

Nor is this pattern confined to a handful of technology firms. Across American industries there is a clear relationship between output growth and price dynamics. The sectors with the fastest growth in real output tend to be the sectors experiencing the largest declines in relative prices.

Computer and electronics manufacturing sits at one end of the distribution, combining exceptionally rapid output growth with annual price declines of around 7%. Much of the rest of the economy looks very different, with slower output growth and steadily rising prices.

Remove tech, and the gap nearly disappears

If the divergence between Europe and the United States is primarily a story about technology and falling relative prices, then the gap should shrink considerably once those sectors are removed from the comparison.

The chart below decomposes real value-added growth into two components, namely the contribution of the technology sector defined as computer and electronics manufacturing plus information and communication services, and the contribution of the rest of the economy.

The United States grew by around 2% per year between 2000 and 2019, but roughly 0.6 percentage points of that growth came from the technology sector alone. Once that contribution is removed, growth in the rest of the American economy falls to around 1.4% per year, almost identical to the Netherlands.

This does not mean that the American economy was no more dynamic than Europe’s. It means that much of the observed divergence is concentrated in a relatively narrow set of sectors rather than spread across the economy as a whole.

That is broadly consistent with Krugman’s argument. The sectors driving much of the American growth advantage are the same sectors characterised by exceptionally rapid output growth and falling relative prices. They contribute disproportionately to constant price measures of growth and productivity, while making a much smaller contribution to measures based on current market values.

At the same time, the chart also highlights that Europe’s performance was far from uniform. The Netherlands looks remarkably similar to the United States once technology is excluded. Italy does not. Its relatively weaker growth performance extends well beyond technology and reflects broader economy-wide productivity problems. In that sense, the debate is not only about the US vs. the EU. It is also about which Europe we are talking about.

The previous section largely supports Krugman's interpretation. But it does not end the argument.

The growth accounting framework allows us to go one step further and ask where productivity growth actually comes from. Is it primarily the result of higher efficiency, better skills, or greater capital accumulation?

The United States does record faster labour productivity growth than most European economies. But what distinguishes the United States is the scale of capital deepening, particularly in intangible and ICT capital. American firms accumulated more ICT and intangible capital than many European economies, although the gap varies considerably across countries and types of intangible assets.

If the American lead reflected primarily higher total factor productivity, the story would be one of superior efficiency or technological dynamism. Instead, much of the difference reflects the accumulation of productive assets that make workers more productive over time.

On its face this sounds like a point for Garicano et al. The US advantage is not simply a consequence of measurement conventions or relative price movements. It has accumulated more of the assets that increasingly underpin productivity growth in modern economies. Workers have access to larger stocks of ICT and intangible capital, including software, intellectual property and organisational assets.

However, a closer look at the ICT component reveals something important. At first sight, ICT capital deepening appears to support the critics’ case. The United States accumulated far more ICT capital services than most European economies, making a significant contribution to its productivity advantage.

But much of this apparent gap reflects the same mechanism discussed earlier.

Measures of ICT capital are constructed in real rather than nominal terms. As the price of computing power falls, a given level of spending translates into a rapidly growing volume of measured capital services.

The United States certainly invested heavily in digital technologies. But the striking growth in measured ICT capital services is not simply a story about spending more money on computers and software. It is also a consequence of the fact that each dollar invested buys progressively more computing power over time as technology prices decline. The same relative-price dynamics that affect measured technology output also amplify measured growth in ICT capital services, since falling computing prices increase the quantity of capital services generated by a given level of spending.

That does not mean the US lead disappears. It does not. But it does suggest part of what appears as an advantage in ICT capital accumulation reflects the same technology and relative-price dynamics discussed earlier rather than an entirely separate source of divergence.

Which Europe?

Once we move beyond ICT capital and look at the broader category of intangible assets, the picture changes somewhat. Organisational capital, training, branding and design are less directly tied to the falling-price dynamics that characterise digital technologies. The United States performs strongly on this measure. Yet once the comparison shifts from US vs. Europe to individual countries, the picture becomes more complicated.

The chart below shows intangible investment as a share of value added across a range of advanced economies. The United States is among the leaders, but it is not alone. France and Sweden invest at comparable levels, while the United Kingdom and the Netherlands are not far behind. The real outliers are Germany, Italy and Spain, which invest less.

Rather than a clean US vs. EU story, the data suggests a growing divergence between a group of more intangible-intensive economies and another group that has been slower to adapt to a more knowledge-intensive growth model.

In that sense, the European productivity debate risks conflating several distinct phenomena. The challenges facing German manufacturing are not necessarily the same as those affecting Italy or Spain. Nor are they identical to the factors behind America’s strength in technology and intangible capital. Aggregating them into a single US vs. EU comparison can obscure as much as it reveals.

There is also a broader measurement issue lurking beneath the surface. National accounts only treat some forms of intangible investment as investment. Research and development, software and databases are included in GDP. Other forms of intangible capital, such as organisational capital, training, branding and design, are still treated as current expenditure even though they contribute to future productive capacity.

In several advanced economies it amounts to between 8% and 14% of value added. The size of this unmeasured capital stock also varies considerably across countries. As a result, international productivity comparisons depend partly on what national accounts classify as investment and what they do not.

That does not invalidate the productivity data. But it does reinforce a broader point running through this debate that measures of output, productivity and capital are not neutral observations of economic reality.

So where does this leave the debate?

Much of the American growth advantage is concentrated in a relatively small number of technology-intensive sectors characterised by rapid output growth and falling relative prices. Remove those sectors and the gap between the United States and several European economies narrows quite substantially. The same price dynamics also help explain why the contribution of ICT capital appears so large in growth-accounting decompositions.

At the same time, the US advantage cannot be reduced entirely to a measurement issue or a consequence of relative prices. The US has accumulated more productive assets, particularly intangible assets, than many advanced economies. That contributes to higher productivity and reflects a genuine increase in productive capacity.

But this is also where the conventional US vs. EU narrative begins to break down. Much of the aggregate transatlantic gap also reflects the weak performance of a subset of European economies rather than a uniform EU failure to adapt to a more intangible-intensive growth model.

The broader lesson is that the debate is often framed as a disagreement about facts when it is really a disagreement about concepts. Living standards, productive capacity and productivity growth are closely related, but they are not the same thing. Krugman is largely correct that Europe’s relative living standards have held up better than many productivity comparisons imply. His critics are largely correct that the United States has developed advantages in the technologies and intangible assets that increasingly drive long-run growth.

Disclaimer: The opinions and analyses expressed here are solely my own and do not represent the views or positions of my employer or any institution I am affiliated with. All content is written in a personal capacity and is unrelated to my ongoing professional work.