Monday, November 30, 2009

Cleaning up a few things

I am bit worried that in my earlier posts I may have come across a bit negative of Nordhaus' project. Although I think with the current methodology it will be hard to design a comparable time series, the project of providing some basic economic geography data is long over due. In doing this it is a remarkable achievment. We know surprisingly little about the global economic landscape and His work is a big start that needs to continue

How that data has been used might be a different matter see World Development Report 2009: Reshaping Economic Geography. goto http://econ.worldbank.org/WBSITE/EXTERNAL/EXTDEC/EXTRESEARCH/EXTWDRS/EXTWDR2009/0,,menuPK:4231145~pagePK:64167702~piPK:64167676~theSitePK:4231059,00.html

If you want to read an interesting review of this report goto:
http://www3.interscience.wiley.com/journal/121683144/abstract Transactions of the Institute of Bristish Geographers (April 2009).

Lastly if you know of relevant economic geography blogs please let me know.

Friday, November 27, 2009

Do networks exclude heirarchies?

Now this post is about an article I don't understand. Recently Peter Taylor has published an article "Urban economics in thrall to Christaller: a misguided search for city hierarchies in external urban relations" Environment and Planning A 2009, volume 41, pages 2550 - 2555 which confuses me.

He opens with:

In my particular intellectual cocoon, I had imagined that the spatial organisation and structure of cities were now generally agreed, crudely, to be a combination of two sets of materialist mechanisms: agglomeration processes creating economic clusters within cities, and connectivity processes creating economic networks between cities. The former has been a very strong research area over recent decades, contrasting with the latter, external urban relations, which has been relatively neglected. By my way of thinking, the two sets of processes are intertwined and between them generate successful, vibrant cities. But, it seems, my cocoon is not particularly encompassing. And it is the focus on networks with their inherent mutuality (horizontal intercity relations) that appears to be far from widely accepted. For some urban scholars, intercity relations can only be understood as hierarchical.

But does a network perspective exclude a hierarchical perspective, are they only merely horizontal?

The article is primarily a critique of the mathematical work of Masahisa Fujita and Jacques-Francios Thisse's (2002) Economics of Agglomeration: Cities, Industrial Location, and Regional Growth (along with the wider work of Krugman) whose agenda is the development of a model city agglomeration and location. There work primarily focusses on central place theory within a nation.

But as Taylor says, Globalisation makes this task more difficult. 'As well as this geographical scalar challenge, their focus on a simple hierarchical spatial structure privileges interurban competitive relations, which means that the inherent subtleties of urban external relations cannot be reached through their urban economics research agenda. Once it is accepted that cities are both competitive and cooperative, then the question arises as to the circumstances in which one intercity relation dominates the other' (p2554).

Then returning to his theme Taylor finishes with 'Breaking free from this pedagogic power, the initial decision in trying to understand cities is whether they are considered to be primarily organised as markets, hierarchies, or networks' .

I don't claim to be an expert in central place theory but what I find interesting about this view of networks is that avoids the issue that even in networks you can have power and hierarchy. Toyata uses a network of suppliers that are co-dependent but Toyota is itself the primary hub - the coordinator and architect. Work on Global Production Networks indicates that in Electronics one company has to be the 'flagship' (Ernst) that takes the lead in design the complex architectiure of the product. Networks don't exclude power relationships they just map them differently.

Anyway I thought this was an interesting recent article that was relevant to the blog. There is probably some nuiance in this paper and as I wonder about it in the earlier work of Taylor that I am missing.

Monday, November 9, 2009

Constant Prices or Share of GDP or ....

The question of how to represent economic structures will be an ogoing discussion in this blog but one of the key aspects of this issue are the methodologies adopted for comparisions.

So for example current prices are virtually useless for anything other than comparing economic phenomena for a single time period for a single nation because it has the same currency. This is because inflation, exchange rates etc cause serious difficulties for anything more than this. You can use calculations like purchasing power parities (PPP) which are better than exchange rates for cross currency comparisons to improve the analysis. By way of example, Nordhaus uses current PPP dollars in his economic geography project. This is fair enough for a single year but once you have a time series this doesn't work.

Constant pricing methodologies improve the utility of time series but they don't capture quality changes or technological changes limiting the benefits;. They are also difficult to use in cross country comparisons.

Typically then, share of GDP has become the standard for international analyses. Publications emerging from organisations such as the OECD almost exclusively use share of national GDP. This is an excellent approach for revealing differing relative specialisations. For example industry X might represent # % in Sweden or ## in the USA. You can even use these comparisions for time series analyses comparing the share of GDP acoss time. This is often done, for example, to highlight the rise of services relative to manufacturing.

But there is a problem. GDP isn't some absolute measure (and there isn't one), obviously GDP can rise and fall and the rates at which it does so differs between economies. The implication is that although this doesn't damage the measure's status as useful for relative specialisations it hides from readers (or blinds them) to the significance of growth. So using our example above in time 1 Sweden's industry represents # % but perhaps the whole economy grows fast but relatively evenly so at time 2 it is still # %. But in comarison to the USA that perhaps grew slowly during the same period - the Swedish industry has actually caught up to some degree on the scale of the industry in the USA.

This is particularly important for our analysis of East Asian economies over the last twenty years and comes to the forefront as we consider how to analyse the BRICS (Brazil, Russia, India, Chin and South Africa) over coming years. We need a complementary measure that captures change better.

What, I have been thinking about is creating a measure based on a basket of economies. Data on the world economy is too problematic, but a basket of economies (say approx 15 or so) where we have good data reaching back to the early 1960s would be workable. Your measure of interest - R&D, industry output etc could then be calculated against this basket of economies. If you wanted to, by the same logic you could create sectoral groups - something that is currently not possible. Over time you would be able to capture the changing dynamics of the world economy. For example, if you created a sectoral group (perhaps auto), in the early 1960s you would probably find that a small group of countries (mostly inside your basket of economies) accounted for a high share of the sectoral bechmark G15 Gross Sectoral Product. However, today, by keeping the basket of economies the same, you would probably find that fewer economies (inside the basket) account for a high share of G15 sectoral product but a significant amount of sectoral product would fall outside the basket.

In this methodology, if a sectoral product basis was calculated we would find that over time "shares" would grow beyond 100%. To me this is not a conceptual but a communication problems in terms of teaching people how to interpret the new measure. For the forseeable future using a basket of GDPs measure wouldn't have this problem, but it will eventually emerge.  We could call the measure something like the Benchmark International Product.

The only genuine challenge to this measure is that of perceived 'western bias'. The first benchmark group would have to have a western' OECD economy bias, which governments in China and East Asia may not like. However if the measure is to be really useful it has to reach back in time to a point where there are only few economies with really good data.