Substack

Wednesday, March 13, 2013

An assessment of "The Economic Complexity Atlas"

I've been thinking of writing about this for sometime. Now that I've passed a preliminary muster of Cesar Hidalgo and Ricardo Hausmann's concept of diversified productive knowledge within an economy, and its role in economic growth, I am left with mixed feelings.

Obviously (or so atleast I think) the Atlas has a two-fold agenda. It has brilliantly explored the manifestations of growth, as reflected in the profile of each country's export products and mapped them. It has also quantified the productive knowledge of an economy in terms of the Economic Complexity Index (ECI), and has shown that this is a reliable predictor of economic growth. Its next objective would be to explore what can be done to help countries acquire a diversified base of productive knowledge. I am not sure whether we are covering much new ground here.

1. The Atlas tells us where countries stand on the complexity map and that complexity appears to be a better predictor of economic growth prospects. So what? Standard endogenous growth theories have for long dwelt on the importance of accumulation of knowledge - both product variety/diversification and specialization - as the critical driver of economic growth.

Yes, the Atlas qualifies the character and dynamics of this knowledge. And Hausmann and Hidalgo believe that this "new" information (about drivers of growth), with its close relationship with growth, would help countries, presumably unaware of this, focus on its acquisition.

But I'm not sure whether there is compelling enough evidence - qualified interpretation of existing theories or new evidence - in anything presented by them to convince political leaders and policy makers anywhere to re-visit their prevailing notions about economic growth. However, it does undoubtedly add significant value to the continuous refinement of the endogenous growth models. But that is a matter of interest for academicians and to this small group sitting in Sweden.

2. Even assuming that we have learnt something new, what does it tell about how to get there? What can knowledge about the current level of a country's productive knowledge base and the desired level of complexity tell about the actual path of getting there? In broad terms, this quest is similar to industrial policy regimes followed by many countries that have sought to promote the acquisition of knowledge and technology in specific sectors.

In fact, the decades experimentation on industrial policy, and its mixed results, should itself caution us about the difficulty of any effort at promoting growth with the productive knowledge destination map. Obviously, even when we know our current location and future destination, we still need to know the path and the vehicles to get there. And if the path is itself a dynamic target and we also have to travel in multiple vehicles at the same time, then the quest for destination becomes even more difficult.

3. It finds that diversity of productive knowledge is a more important predictor of growth than even investments in human capital. Maybe. But even controlling for various "observable" factors, we simply cannot quite infer causal relationship between acquisition of diversified productive knowledge and economic growth. The interaction dynamics of the contextual (I am wary of using "institutional"!) factors are too complex to do that with any degree of reliability. It could just be that countries with high ECI scores also had the right environments to acquire that knowledge.

4. Then there is the practical difficulty of its assumptions. Clearly, we cannot expect the numerous small least developed countries like Gabon, Sudan, and Malawi, who are languishing at the bottom of the ECI rankings to do anything much, even in the medium to long-run, to acquire a diversified base of productive knowledge. In an ideal world, the inhabitants of these places should merely migrate to other more habitable locations. In the real world though, their most realistic chance of expedited development lies not in acquiring diversified productive knowledge, but in being able to achieve success with a few sectors.

5. The Atlas points to the relatively low growth potential of the richer natural resource endowed economies, attributing it to their low productive knowledge base. But it can also be argued that given the low economic base from which they started, the natural resources provided them to best and fastest path to economic growth. And atleast some of them grew faster than their other similarly placed colleagues. It is true that many of them have not leveraged that growth to diversify their economies (and expand their productive knowledge), which in turn raises questions about the sustainability of their growth.

While we may claim that once we control for the role of oil, the Middle Eastern economies, with their low productive knowledge base, would have been as backward as any other developing country, it cannot be denied that the people of these countries today enjoy lives which are far better than those in non-resource developing countries, including those with much higher productive knowledge base. If they, like Norway, had used resource windfall to build up a modern economy, we would today have been talking about that as a growth model. There is no reason to lend any more credibility to the productive knowledge base model than one which advocates leveraging natural resources to develop an economic base and then move up the technology frontier.

6. Finally, the Atlas is created using export data. This may be a less reliable proxy for the productive knowledge base of the larger economies, whose domestic economy may contain a diversity of productive knowledge that may not be appropriately reflected in the exports.

Lest we need any more reminders, all "holy grail" explanations of economic growth and development are futile exercises. It all ultimately boils down to exercises in circular logic. Critics will rightly point out that countries lagging behind on diversified productive knowledge base do so because of lack of effective institutions, culture, inadequate investments in human capital, geography, historical factors, weak governance systems, and a series of other factors. Supporters will say that the process of acquisition of productive knowledge will help overcome these other deficiencies. And so on...

3 comments:

Martin Mattsson said...

I don't agree that any attempt to explore what factors have been important in historic growth experiences is futile. However, I do agree with your point no. 3. There seem to be very little empirical evidence that the story is causal and not driven by some other underlying factor. Furthermore, even if complexity is what drives economic growth it is not clear that employing policies that are aiming to increasing complexity is a good idea (this is analogous to saying that since we know that capital accumulation is a proximate cause for growth we should force low-income countries to save more).

Unknown said...

I find point 2 as the main question and thus contribution of this blog. Indeed, the reverse engineering of the ECI through viable policies and other interventions/instruments is the key quest, and any contribution in that direction is much welcome.
Should the latter issue get resolved and then applied (which will take some time!) will we be able to evaluate the pertinence and novelty of the ECI.
This last point should also answer, to some extent, the concern of the previous comment.

Mark Belcher said...

Thank you for this thoughtful blog post and the associated comments.

I share many of the concerns raised here, but also believe there's practical benefit to be gained from this type of analysis. For instance, the "Export Opportunity Spectrum" (EOS) that was calculated could be of potential benefit to development agencies and practitioners in answering the central question of how and where to invest resources to promote export-led economic growth.

Point 4 above is a valid observation but I am not sure that is what the authors are arguing. The image they use (rather unfortunately perhaps) is of primates inhabiting a forest. Product groups (trees) that are "closer" in terms of complexity are easier to populate by jumping to the next closest tree. The implication is that incremental moves up the complexity scale are preferable to "great leaps" that would be more costly to achieve and have a higher potential rate of failure.

EOS seems to offer a ready-made (i.e., free) quantitative basis for identifying product groups that are within "reach" in terms of a country's current level of complexity and that offer the greatest potential development payoff by capitalizing on under-utilized Revealed Comparative Advantage.

Using EOS as a targeting tool for conducting deeper analyses at the value chain (VC) level is where I think the utility comes into focus. You correctly note that there are a variety of context-specific variables (e.g., laws, policies, institutions, macroeconomic conditions, market structure, VC- and firm governance, workforce capacity, social exclusion, etc.) that have to be understood and addressed in any given case. EOS doesn't help at that level.

So, it seems that there may be three potential benefits of using EOS as suggested above: 1) it helps avoid the pitfall of "boiling the sea" to identify unrealized export potential across an entire economy; 2) it provides a data-driven basis for making more nuanced decisions about where to invest donor resources to promote export development; and 3) it provides a useful quantitative baseline against which development impact can be measured.