Scenarios of technological progress in Italy: what can we expect?

ABSTRACT

The simple but deep sense of technological progress (TP) lies in the possibility of improving human life. The immediate question thereafter is clearly about the distribution of the gains from TP. With the diffusion of automation and artificial intelligence, fears about the implications of TP on employment and wages have gained renewed importance. While scholars are divided on the effects of this new wave of TP, they agree that every economy will be affected differently, and hence, it will require tailored policy measures. In this paper, we frame how TP could affect the Italian economy, as it is now. We simulate four different scenarios through a dynamic computable general equilibrium model with three types of labour and six types of households. We calibrate the model on the social accounting matrix, and we find that TP returns higher growth patterns albeit with disruptive effects on labour.

KEYWORDS:

• Automation and robotisation
• skills and employment
• labour demand
• Dynamic CGE model
• social accounting matrix

JEL:

• C68
• D58
• E16
• E24
• J24

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/13662716.2022.2152313.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ Keynes (1939) did not prevent Bennion (1943) to revamp his wording later, however.

² Indeed, what makes automation attractive is its cost-saving feature.

³ Jestl (2022), for example, provides a brief but interesting overview of these papers, which he groups according to the aggregation level (sectoral, firm, regional) adopted in the analysis.

⁴ In Zotti et al. (2020), the SAM represented the database for the Macro Multiplier methodology, as developed in Ciaschini and Socci (2007).

⁵ See Tables A1 and A2 in appendix A for the list of commodities and activities.

⁶ This disaggregation is carried out using data from the OECD Programme for the International Assessment of Adult Competencies (PIAAC) since it is compatible and complementary to the ISTAT data on digital competences.

⁷ This index is obtained as a weighted average of the use of email (5%), the use of internet to better understand questions related to work (7%), e-commerce and e-government (10%), the use of MS-Excel (13.5%), the use of MS-Word (17.5%), the use of complex programming (22%) and the use of video-conference and other real-time participation to discussion (25%). For this disaggregation, we have used the data from PIAAC database integrated by EU-SILC.

⁸ See https://www.siope.it/Siope/.

⁹ See https://www.istat.it/it/archivio/204387.

¹⁰ The complete formulation of the model is presented in the Appendix B.

¹¹ See Appendix B for a list of parameters employed in the model.

¹² For this motivation, we considered these parameters robust and did not include a section dedicated to the sensitivity analysis.

¹³ Since there is no relevant literature nor available which can corroborate any assumption about the substitutability between these types of workers, it is necessary to resort to the case of either perfect substitutability or perfect complementarity. Both these assumptions allow avoiding the typical problems related to the introduction of exogenous parameters (i.e. sensitivity analysis and robustness checks, see Grassini 2009). For this reason, we assumed the complementarity in these nests, as the worst possible scenario that encompasses an immobility of the labour market for the above-mentioned typologies of workers, and focus on the substitutability among low-medium-skilled occupations, that show more contributions in the literature. This might raise question about the opportunity to have such a deep distinction in workers characteristics. The rational, behind such a deep disaggregation is related to the potential information that can be collected from the simulation results in disaggregate terms.

¹⁴ Following the Armington assumption, consumers are able to perceive the differences between imported and domestic goods even if they belong to the same category. Therefore, we assume that domestic and imported goods are imperfect substitutable and the elasticity of substitution is positive as in Zotti et al. (2020).

¹⁵ We consider the small open economy assumption.

¹⁶ The model replicates in the first period the flows of the SAM and in the subsequent periods it reproduces the steady state growth path determined assuming exogenous depreciation rate, interest rate and total output growth rate that are compatible with the level of investment and the initial stock of capital derived from the SAM (Lau, Pahlke, and Rutherford 2002).

¹⁷ Over the 2011–2020 period, the average yearly growth rate of productivity has been 0.9%.

¹⁸ Among others, this literature includes the studies of Frey and Osborne (2013, 2017), Brynjolfsson and McAfee (2014), Ford (2015), World Development Report (2016), McKinsey Global Institute (2017), Arntz, Gregory, and Zierahn (2016; 2017), Nedelkoska and Quintini (2018).

¹⁹ Even one very recent study (i.e. Jestl, 2022 that dates June 2022) focuses on the period 2001–2016.