Guerrieri and Paodoan: Modelling ICT as a GPT

Guerrieri, Paolo, and Pier Carlo Paodoan. 2007. Modelling ICT as a General Purpose Technology: Evaluation Models and Tools for Assessment of Innovation and Sustainable Development at the EU Level. Collegium no. 35.

This report was done in an attempt to further the Lisbon Agenda, specifically emphasizing the improved usage of ICT. “Most of the existing models linking ICT to economic growth and employment, both at the macro and the micro level, do not provide a fully satisfactory analysis of the transmission mechanism of ICT to economic performance and do not take fully into account the response of different national systems of production and organization, including the role of business services, to the development and diffusion of the new technologies” (6).

“More progress is also needed in the analysis and empirical investigation of the determinants of ICT spending, which is largely if not completely determined by the private sector” (6).

“Finally, data limitations and the lack of appropriate sets of indicators have been a severe constraint in modeling the impact of ICT related policies” (7).

General Purpose Technologies are: pervasive, technologically dynamic, exponentially impacts productivity (7). ICTs are a GPT. Three ways that it impacts the economy: ICT production; ICT as capital input (importance of computers and information technology as an input in other industries); ICT as special capital input (positive externalities resulting form ICT investment) (8).

Many of the effects of ICT investment are indirect and long-term (8).

Data: Best source is EUROSTAT, especially for human capital information. Time series short. GGDC is a good source for ICT variables and indicators, come from national stats, etc. OECD data have many missing variables, data points.

A “Core Dataset” is then created using a hybrid of GGDC and STAN OECD numbers.

Use 8 CGE models to explore ICT. Find IFs to be quite useful, as it is the only model that explicitly incorporates an ICT sector.

“We have chosen the IFs model to perform simulation exercises, since it includes a specific ICT sector and it allows CIT to exert its impact on the economy via different channels, rather than being modeled as a simple input in a standard production function” (13). They consider a total of 6 different scenarios (13).

A list of their results can be found (i-vi) on p. 14.

“A report by Indepen (2005) underlined that simply increasing total investment in ICT will, in itself, not deliver improvements in productivity and economic growth. To be productive, this investment also requires complementary changes in the way organizations are structured and function as well as in human capital” (15). There is a list of 5 of these different catalyzing factors on page 15.

“Modelling ICT as a GPT requires a multi step procedure” (19):
-when related to TFP, relationship between GPT and ICT; relationship between ICT and economic structure and relationship between ICT and facilitating environment/structure
-these elements interact in a virtuous cycle

Ch. II: ICT as a General Purpose Technology

This chapter is a literature review that situates ICT within the GPT literature, as well as presenting a framework for thinking about a modified production function with a ICT spill-over effects being represented.

“One of the fundamental insights provided by Schumpter (1939) is that technological innovations are not evenly distributed over countries, industries and time” (22). This is then deployed by Perez (mostly) and Freeman to explain techno-economic paradigms. This is explained in depth on 23-4.

GPT: “GPTs are radical new ideas or techniques that have the potential for important impacts on many industries in an economy” (24).

Techno-economic paradigms are compared to GPTs by Helpman (1998). Helpman concludes that, “…the concept of TEP is much broader than that of GPT since it covers the entire economic system surrounding any set of pervasive technologies actually in use” (24).

ICT -> economic production: “…innovation is a complex phenomenon that goes beyond an increase in R&D expenditures and whose impact cannot be understood in the framework of a production function” (25).

A neoclassical aggregate production function is compared with a Structuralist-Evolutionary (S-E) decomposition. The later incorporates a more nuanced and structural account of technology mediated through policy environment affecting productivity and altering performance (26).

“In particular our aim is to shed some light into the black box of the production function and to capture the interactions between ICT, the facilitating structure, public policy and performance” (29). Their assumptions: ICT impact depends on nature of environment; facilitating structure comprised of human capital, regulation and sectoral composition of an economy; certain economic sectors have more impact; R&D is crucial; and ICTs are pervasive technologies (29).

ICT affects production, according to Schreyer (2000) in three ways: production, capital input and special input. The most important influenced occurs through spill-over effects. These effects the authors attempt to replicate through a production function. “In fact spillover…effects of ICT on GDP have been shown to be larger than direct effects” (39).

Ch. III: Information Society and E-Inclusion: Data and Indicators

The goal of this chapter is to overview different data sources and create a core dataset. “Another goal is to explore available data in order to asses which type of information can be ‘consistently’ used in the implementation of multivariate studies on ICT diffusion” (40).

”In the literature on economic growth three effects of ICT on productivity and growth are distinguished. First, ICT investment contributes to overall capital depending thus helping to raise labour productivity. Second, fast technological progress in the production of ICT goods and services may contribute to more rapid multifactor productivity (MFP)_ growth in the ICT producing sector. Third, larger use of ICT may help firms increase their overall efficiency and thus raise MFP” (41).

The following data bases were screened: “…Total Economy Growth Accounting and the 60-inidustry database of the Groningen Growth and Development Center (GGDC); the OECT STAN database for Industrial Analysis and the OECT Main Science and Technology Indicators (MSTI); EUROSTAT, R&D and National Accounts database…In principle, the best coverage in terms of variables/ indicators and countries is given by EUROSTAT data. Time-series length however is too often below 10 years…” (43).

Data sources evaluated on 44: the best overall source is EUROSTAT, especially for human capital and policy variables, however, time series are too short; GGDC is the best for ICT variables and indicators; OECT data have many holes, but are good and straight forward (44).

The authors then develop a “Core Data Set” which is a combination of both OECD and GGDC numbers.

There is some statistical analysis of the Core Data Set, and one of the main findings is a strong relationship between IT expenditures and regulations. More detailed analysis of the results on page 56.

A discussion of e-inclusions follows, with its definition and the reason that it is an important factor for European development. These discussions revolve around the literature on the digital divide. This then moves into discussions of digital literacy, how e-inclusion motivates and empowers, ehealth, egovernment, e-everything.

E-inclusion variables are listed in a table on 75-78.

Ch IV: Technology and Performance

“The purpose of this chapter is twofold. On the one hand, the evaluation will be used to choose the CGE model to be employed for simulation exercises of the impact of ICT on economic performance; on the other hand it will help identifying [sic] possible ways of introducing ICT in several CGE models…” (83).

Heavy focus on how ICT can be modeled through endogenous productivity in a CGE with an emphasis on spill-over effects. Definition of ICT: “…the OECD provides an official definition of ICT, and ICT goods and services are evaluated taking into account their ‘intrinsic‘ quality” (84). The authors also defined productivity as relating to TFP contributions by the ICT sectors. They explain that it is easier to identify positive productivity contributions made by ICT in the production sector, but much more difficult in the service sector. They also identify different types of time lag patterns associated with different structural factors within countries that make them more or less likely to reap the rewards of improved and increased use of ICT.

They then compare 8 models: NEMISIS; MULTIMOD; WORLDSCAN; QUEST; NiGEM; IFs; Oxford World Macroeconomic Model and GEM E-3. They felt like IFs was quite suitable for their purposes because it has a thick endogenous technology model as well as an explicit ICT sector, as well as an ability to alter productivity through R&D spending.

They also claim that IFs does not take into consideration technological spillover effects (102).

They then run simulations with many of the models and explore the various assumptions of each model.

Ch V: ICT as a General Purpose Technology

“The existing CGE models with endogenous technical change cannot fully capture tehh complex mechanisms through which ICT influences economic performance….can be improved in three ways…First, the list of endogenous variables that are affected by ICT could be extended by including employment and international competitiveness. Second, the interaction between ICT and other variables affecting the overall business environment (such as regulation, sectoral composition, and human capital) could be taken into account. Third, the uneven impact of ICT across sectors could be considered as well” (115). Also, claim that a higher degree of sectoral disaggregating would make it easier to model the effects of ICT.

They want to explore a variety of variables relating to ICT and see how this effects productivity differently. They run scenarios with IFs that firstly look at variables that more directly drive production, and then at diffusion effects (119).

There is a review of the structure of the IFs model generally, with a focus on the drivers of MFP.

Scenario A: The role of capital accumulation in the ICT sector. “This scenario aims at understanding the effect of a more productive capital factor in the ICT sector” (124). Conclusion: there is a long time lag.

Scenario B: The diffusion process of ICT best practices across sectors. “The scenario tries to evaluate the impact of a more rapid rate of adoption and diffusion of ICT through the service sector, which is typically a large user of ICT” (128).

Scenario C: The diffusion of ICT across countries. “In this scenario we assume different regimes for the MFP of the ICT sector for technological leaders, in order to evaluate the diffusion of different technological frontiers reached by a country which is assumed to be a system leader” (132).

Scenario D: The effects of past adoptions of ICT. “In this scenario we simulate different regimes regarding the adoption of ICT technologies in the economic system” (137). This scenario focuses heavily on the number of networked people variable in IFs.

Scenario E: The role of institutional setup I. “..effects of a change in the elasticity of the multifactor productivity of the ICT sector to the level of network infrastructure in the economic system” (141).

Scenario F: The role of institutional setup II. “…change the value of the elasticity of MFP of the ICT sector to the infrastructures in communication technology” (143).

Ch. VI: Endogenizing ICT

Exploration of the digital divide and literature explaining the phenomena with different drivers highlighted: Income? Human Capital? Regulation? Demographics? Economic Freedom? Service/Agriculture sectors of the economy? Price index?

“Our empirical results are in line with the literature on the digital divide…Human capital, and investments in R&D are factors that increase ICT investments, while burdensome regulation tends to depress them. Furthermore, the structure of the economy turns out to be a relevant factor to understand the different rate of investment in ICT; in particular, countries with a higher share of the service sector usually display hither ICT investment” (156).