Somebody FINALLY Writes A Decent Paper On The Middle Income Trap

I’ve been meaning to cover this since it landed in my inbox last week, but hadn’t found the time. But this new IMF working paper gets a big thumbs up from me (abstract):

Growth Slowdowns and the Middle-Income Trap
Aiyar, Shekhar and Duval, Romain; Puy, Damien; Wu, Yiqun & Zhang, Longmei

Summary: The “middle-income trap” is the phenomenon of hitherto rapidly growing economies stagnating at middle-income levels and failing to graduate into the ranks of high-income countries. In this study we examine the middle-income trap as a special case of growth slowdowns, which are identified as large sudden and sustained deviations from the growth path predicted by a basic conditional convergence framework. We then examine their determinants by means of probit regressions, looking into the role of institutions, demography, infrastructure, the macroeconomic environment, output structure and trade structure. Two variants of Bayesian Model Averaging are used as robustness checks. The results—including some that indeed speak to the special status of middle-income countries—are then used to derive policy implications, with a particular focus on Asian economies.

I’ve generally found academic research into the middle income trap (I refuse to dignify the expression with capital letters) to be largely unsatisfying. Either these papers write about how to get out of one and take the phenomenon as given or when investigating it, use atheoretic arbitrary definitions like the number of years taken to exceed a certain income level.

I find the latter variety more than a little hard to accept, because you can spin the results any way you want by tweaking the income levels or the number of years used i.e. it’s open to fitting the data to the model rather than the other way around. In other words, you can easily “prove” any particular preconceptions or stylised facts you may hold, which isn’t rigorous enough proof for my taste.

The other thing of course, is by any of the definitions I’ve seen in use, nearly every single developed country in the world today has been in a middle income “trap”, simply by virtue of the length of time taken between income levels (e.g. Britain during the Industrial Revolution).

But this paper is DIFFERENT, and those capitals are deserved. The biggest difference is that the link between growth and development theory with the empirical data is finally there. The authors define the middle income trap very specifically as deviations from convergence, not as slowdowns in growth per se. That might not seem a big deal, but it effectively removes much of the potential for data mining from the econometrics.

Let me explain: the idea of convergence springs from the basic neo-classical growth model. Growth is derived from factor inputs (capital and labour) and technology. As countries develop, growth in factor inputs slow – greater capital stocks attract greater risk of obsolescence, requiring higher rates of investment to maintain a given level of “productive” capital stock. Population growth first accelerates as lifespans increase and infant mortality drops, then decelerates as fertility later drops.

In the end productivity becomes the final arbiter of economic growth. The quintessential middle income trap occurs when productivity growth in a middle income country is insufficient to maintain progress towards high income levels when the growth in capital and labour has already slowed.

Technology has a special role to play here as developing countries, by definition, have a lower technology level than the most advanced economies. Growth can thus be accelerated by “borrowing” technological advancements from other countries. As the technological level starts approaching the global ceiling, this source of advantage will slowly disappear, and any further growth needs to come from innovation, i.e. internal additions to the level of technology.

The upshot of it all is that growth slowdowns are a “natural” occurrence, a consequence of declines in the increase of factor inputs as well as a technological level that is approaching the global maximum. Barring setbacks, a country on the development path should see its income level “converge” very quickly towards the global maximum when beginning from the bottom, but gradually slow as it reaches the heights.

With this in mind, the middle income trap thus would not be a growth slowdown in absolute terms, but a slowdown relative to the growth trajectory defined by convergence, conditional on existing factor endowments.

Put more simply still – imagine a road that goes up, and gets progressively steeper as you climb. The stronger you are the faster you can run to the top, but the higher you climb the harder it is to keep going at the same speed.

You’re only in an income trap if you’ve gotten off the road. This could be from tiredness (low productivity), but it could also be from tripping up (policy mistakes), or someone pushing you off (war).

Theory and empirics coming together – I love it!

A second difference is that the methodology used allows for testing income traps at all income levels, not just in the middle income strata. And they test for multiple causal factors of convergence slowdowns (both in levels and in differences), checked robustness through different income definitions – as I said, I love this paper, they touched so many bases.

As for the results – they come to the conclusion that a middle income trap exists (higher propensity for middle income countries to experience convergence slowdowns), that convergence slowdowns are a higher risk for developing countries, and that episodes of convergence slowdowns peaked in the 1980s (25% of the sampled countries). The latest data for 2000-2005 shows incidence of convergence slowdowns had dropped to just 9%.

Probably more important than the identification of convergence slowdowns, are the policy implications. Page 29 has a summary table of the results (for policy makers: the heatmaps on Pg 33 –34 would also be very informative, as is the spider web analysis on Pg 36). The table basically outlines those reforms and policy initiatives required to avoid a convergence slowdown.

There are some interesting quirks here – for example, deregulation helps to avoid slowdowns, but once regulation is already light, further deregulation does not matter. Same thing with government size – shrinking the government sector helps, but only if you keep doing it. Since there’s a limit to downsizing, that suggests that you can only reduce risks just so much. Strong rule of law however, always matters. High capital inflows also matter a great deal – capital controls, anyone?

And so on to the trillion dollar question – is Malaysia in a middle income trap? At least as defined within the sample period in this study?

And the answer is…

…No, but we used to be.

Convergence slowdowns in Malaysia occurred in two periods: 1980-85 and 1995-2000. These episodes coincided with deep recessions i.e. not the sustained slower growth paradigm of a “traditional” middle income trap. In the former, Malaysia was the odd man out along with Hong Kong among a gaggle of Middle Eastern and Latin American slowdowns. The latter episode, during the Asian Financial Crisis, was more a regional phenomenon.

One last thing of note here – even with its firmer theoretical foundations, this paper also basically rebuts the basic tenets of the middle income trap hypothesis as it is typically defined. Growth slowdowns across the globe, of both the absolute and convergence variety, happen for a variety of causes not necessarily directly or even indirectly linked with productivity.

In the end what it all boils down to is the specific drivers of growth, which would be different for each country with their different levels of capital, labour and technology, geographical locations, cultural and historical contexts, and a myriad other variables.

Technical Notes:

Aiyar, Shekhar and Duval, Romain; Puy, Damien; Wu, Yiqun & Zhang, Longmei, "Growth Slowdowns and the Middle-Income Trap", IMF Working Paper No 13/71, March 2013