World-System (1960-2100) WL20 Technology Growth Paths




The 2025 Nobel Prize in Economics was awarded for both advances in our theoretical and historical understanding of Technological Change. One reading of the work by Joel Mokyr, Philippe Aghion and Peter Hewitt is that new discoveries created by the random historical process of invention will always provide the technologies (through Creative Destruction) necessary for continued, exponential economic growth. 

The forecast above, generated by different WL20 Models, shows that emphasis on Technical Efficiency (TECHE), particularly reducing Emission Intensity (CO2/EG), provides the best future growth path for the World System. Economic Theory seems to be mostly focused on Technical Productivity (TECHP).

The historical work for this Cornucopian conclusion about the Future (Joel Mokyr) was based on studying the Industrial Revolution during the period 1500-1700. The theoretical work (Philippe Aghion and Peter Hewitt) was based on a formal mathematical economic model that, although it could not directly be tested, made predictions seeming to be confirmed in current Western European Countries.

General Systems Theory and World-Systems Theory have a lot of criticisms that can be leveled at the historical and theoretical approach underlying the 2025 Nobel Prize in Economics, criticism I'll discuss in future posts. For this post, I will just present how Systems Theory (1) defines Technology (it is somewhat different the the Economists definition) and (2) would generate an estimable statistical model of Technological Change for the World-System.

Systems Theory models do not generate unending exponential growth as a typical model although unstable growth is possible if the models are destabilized.

System Theory identifies four types of Technological Change models: (1) Technical Productivity (TECHP, output per person), (2) Technical Efficiency (TECHE, output per input), (3) Learning by Doing (endogenous change) and (4) Random Walk (RW) Technology (inventions appear randomly over time). The four models are tested separately. The best model for a given historical period is identified using the Akaike Information Criterion (AIC). The models are displayed in the Notes below.

The best short-term, year-to-year WL20_TECH model is the Random Walk (RW). The best long-term Attractor Path model is the TECHE model. The WL20_TECH BAU model with Learning by Doing produces growth-and-collapse starting after 2040. The Philippe Aghion and Peter Hewitt model resolves to a random walk with drift. 

As a technical exercise, you can modify the WL20_TECHE BAU Model (here) to a RW with drift by changing the diagonal elements of the BAU System matrix to unity (F[1:3,1:3] = 1.0). 

Also notice in the TECHE Measurement Matrix that TECH1 is an historical controller balancing growth in efficiency against CO2/EG efficiency. Another way to look at this results is that CO2 emissions decline over time for technical efficiency to increase.

Technology also might not be the most important driver of Economic Growth. Hegemonic Leadership may be equally or more important and may in fact be a driver of Technical Change (see the Hegemonic Dominance forecasts here).

You can run the WL20_TECHE Model here. Instructions for creating unstable, unending, exponential technological change are given in the code.

Notes

All the measurement matrices were estimated from World Bank Data using Principal Components Analysis (PCA)--see the Boiler Plate for more information. All Systems models were estimated using the dse package in the R programming language.

AIC Statistics

TECHE Model System Matrix



Notice in the WL20 TECHE model that unending, unstable, exponential growth (F[1,1] > 1.0) is improbable given the bootstrap confidence intervals (CI) where F[1,1] = 0.87076812 and the CI does not bracket unity.


WL20 TECHP Measurement Model


TECHP1 = 0.4700 (CO2/N) + 0.510 (EG/N) + 0.505 (GDP/N) + 0.514 (L/N)

WL20 TECHE Measurement Model


TECHE1 = 0.607 (L/N) + 0.554 (GDP/L) + 0.119 (EG/GDP) - 0.557 (CO2/EG)

WL20 BAU System Matrix

The Learning-by-Doing (endogenous change) coefficient is 1.01169571, F[1,1] in the Systems Matrix of the BAU model.

WL20_TECHE System Matrix



WL20 RW System Matrix












Comments

Post a Comment

Popular posts from this blog

World-System (1950-2150) Stable French Technology Cycles

Image
   The 2025 Nobel Prize in Economics ( here ) was awarded for having explained the role of innovation and  Creative Destruction  in  Sustained Economic Growth . Particularly interesting (to me) was the Aghion and Hewitt paper   A Model of Growth Through Creative Destruction , which won them part of the Prize. Does the 2025 Nobel Prize have anything to say about current French economic development and political instability? I have two explicit models for technological change, one for Productivity change ( TECHP ) and another for Efficiency change ( TECHE ). The models are based on  State Space  indexes (see the Notes below) constructed using  Principal Components Analysis  ( PCA ). The  Phase Space  for the FR TECHE model is displayed above. The  Phase Space  for the FR TECHP model is displayed below. Both models are cyclical and stable with periods of over a Century (see the  Notes  ...
Read more

About

Image
  This blog documents an approach to Technological Change that is consistent with General Systems Theory, which contains models that cover Neoclassical Growth Technology, Learning-by-Doing Technology, World-systems Theory and Classical models of Technology to include Schumpeter's Creative Destruction. None of these models are consistent over time within a single country and across countries. Partly, the changing approaches to Technology are historical and partly they are due to data limitations that prevent consistent testing over all time periods and countries.  Therefore, there is no general conclusion about technology accept that it may follow one of the time patterns above (unlimited exponential growth, growth and decline or decline). Moreover, Technological change is not always the best model to describe growth across time and countries. Different models may dominate during particular periods and in particular countries.
Read more