Middle income trap – it may exist but it may not matter too much


May 1, 2024

There is some evidence suggesting that catching up becomes more challenging when a country reaches relative productivity levels of about two-thirds of the USA. However, this still implies growing GDP per capita.

In our previous post we wrote about our simple growth model and projections far into the future (link). In this post we would like to focus on productivity growth, which is a crucial element of long-term economic growth.

Traditionally, growth models assume a convergence phenomenon: the lower a country's relative productivity, the faster its growth rate should be. The idea is that implementing technologies and practices developed elsewhere should facilitate growth. Broadly speaking, the evidence aligns with this theoretical assumption, albeit with considerable variability. While some countries exhibit low productivity growth despite their low income levels, (see chart below) fitted curves based on growth experiences generally support the relationship on average (as indicated by the blue line). Additionally, they suggest the possibility of a "middle income trap" at around 70% of USA productivity.

The notion of a "middle income trap" posits that economic convergence halts at a certain level of development. (see for example: https://documents1.worldbank.org/curated/en/965511468194956837/pdf/104230-BRI-Policy-1.pdf ). Convergence is relatively straightforward at lower income levels, but as countries approach higher income levels, importing existing technology may become more challenging. There's an ongoing debate about whether this phenomenon truly exists or if economic convergence is simply slow and uneven.

The chart below illustrates the initial relative level of Total Factor Productivity (TFP) on the horizontal axis, compared to the USA, and average growth in the next 10 years on the vertical axis. This time frame should mitigate the influence of cyclical fluctuations, as most countries are likely to experience a full business cycle during this period.

The blue line on the chart represents a logarithmic fit of the starting relative TFP, with GDP size used as a weighting factor to prevent skewing by small economies with unusual experiences. This shows a consistent decline – the richer the country, the slower the TFP growth. However, this approach imposes uniform parameters across the entire range of initial conditions, resulting in a linear relationship in terms of log GDP (illustrated by the smooth curved blue line).

To explore whether a local slowdown pattern this exist in our data, we fitted lines that allow for some “wobbliness” by analyzing the data locally, using a moving window. The red and green lines represent these fits (equally weighted and GDP weighted, respectively). 

Tax havens/financial centers and economies with large natural resource revenues are omitted.

Notably, there appears to be no TFP growth on average at around 70% of the US level, based on local smoothing, while positive TFP growth is evident at levels around 50% and 90%. This suggests the presence of a higher-middle-income trap! However, it's crucial to acknowledge that this pattern may not persist in the future, and even historically, it did not necessarily translate to zero per capita GDP growth on average, as increasing employment and capital stock partially compensated for the lack of productivity increase. It's also essential to recognize that we're discussing local averages, around which there is significant variation in country experiences. In other words, the "trap" does not apply universally to all countries.

Let’s assume that this pattern of middle income trap holds in the future. What would that mean? In that case, countries below approximately 70% of US productivity would gradually regress to a level where their growth rates equal those of the USA (and other “high equilibrium” countries). Let’s say this level is roughly 50% of the USA. Conversely, countries above a critical level would eventually converge to US productivity. Ultimately, there would be two "focal" levels of relative productivity – around 50% and 100% compared to the USA – around which countries would cluster in terms of productivity. This assumes no leapfrogging/jumping over the danger zone at around 70% productivity level. 

There may really be a tendency for multiple focal points for development, the concept of "convergence clubs" is gaining traction in economic research. Yet relatively low productivity levels can be offset by a larger stock of capital in the Solow model and increased employment relative to the population. Therefore, being stuck at this relatively low level does not necessarily entail proportionally lower living standards in these countries. Moreover, GDP growth can remain positive even without these factors, as long as the productivity frontier (in this case, the USA) is advancing. Thus, even countries trapped in this scenario would eventually become wealthier, albeit not in comparison to the USA and in the end the existence of this “trap” would not matter that much.

In real-world scenarios, various additional factors come into play, including economic policies, efforts to enhance human capital, and innovation. Even if the middle-income trap does exist, it is likely not deterministic, and its application probably varies across countries at a given level of relative productivity.