The impact of ultra-fast broadband on labor income: an event study approach

Abstract

ABSTRACTWe investigate the impact of ultra-fast broadband connections on labor income and employment. We use panel data for Italian municipalities for the period 2012–2019 and we exploit the staggered roll-out of ultra-fast broadband started in 2015. Through an event study approach, we find evidence of endogeneity between ultra-fast broadband roll-out and labor market outcomes. To identify causal relationships, we use income from pensions to implement the estimator developed by [Freyaldenhoven, S., C. Hansen, and J. M. Shapiro. 2019. “Pre-Event Trends in the Panel Event-Study Design.” American Economic Review 109 (9): 3307–3338. https://doi.org/10.1257/aer.20180609.]. We find that access to ultra-fast broadband increases the income of the self-employed by 1.3% but has no impact on workers. Such an effect is mostly driven by a rise in self-employed workers, which is concentrated in urban areas, and in municipalities at the top and bottom quartiles of labor income.KEYWORDS: Ultra-fast broadbandfiber-based networkslabor incomeself-employed workersJEL CODES: L96D24D22 AcknowledgmentsWe would like to thank the Editor, three anonymous Referees, as well as Fabio Landini, Giovanni Cerulli and the participants to the SIE 2022 (Torino) and SIEPI 2022 (L'Aquila) for useful comments and suggestions to previous versions of the paper. We are grateful to Mario Mirabelli (TIM-LAB) and Francesco Nonno (OpenFiber) for providing us with access to and guidance on the broadband data used in this paper. The views expressed herein represent those of the authors and do not reflect in any case the opinions of the companies and institutions that provided the data and funding.Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1 Indeed, starting in 2018, the Italian government has increased the financial resources from 0.5 to 7 billion Euros for UBB. In 2021, the Italian Government has decided to use part of the Next Generation EU funds to finalize the deployment of UBB infrastructure throughout the country, with around 3.6 billion Euros of public expenditure.2 The two papers also differ in the UBB variable used. While we consider a dummy variable describing the availability of a UBB access in a municipality in a given year, Abrardi and Sabatino (Citation2023) use the number of years since UBB was introduced in a given municipality.3 Higher broadband speed levels may also affect property prices (Ahlfeldt, Koutroumpis, and Valletti Citation2017) and firms' location decisions (Canzian, Poy, and Schüller Citation2019; Duvivier Citation2019).4 The Digital Agenda for Europe specifies the goals in terms of network coverage and service adoption for the whole European population. See https://www.europarl.europa.eu/factsheets/en/sheet/64/digital-agenda-for-europe for more.5 https://www.agcom.it/documents/10179/1571667/Documento+generico+08-11-2014+1415441917492/d34cc914-c150-4fd7-a383-a0c39c9d7670?version=1.16 Before 2015 only a few large cities such as Milan and Bologna enjoyed fiber-based connections realized by the local telecommunication operator.7 Open Fiber deployment plan can be found here: https://openfiber.it/area-infratel/piano-copertura/.8 For privacy reasons, data are missing when municipalities have less than three taxpayers for a particular category of income. This explains the lower number of observations for self-employed income, as in small municipalities there may be less than three self-employed workers. For the calculation of total labor income, we treat missing values as zeroes.9 Results are not affected by different clustering methods.10 Since our sample covers from 2012 to 2019, then r={−7,−6,…,0,+1,..,+4}.11 In Italy, the pension benefit is indexed to the accumulated lifelong contributions valorized with the nominal GDP growth rate (as a five-year moving average).12 The Italian government introduced some (limited) flexibility only after 2019, by allowing early retirement under specific age and contribution conditions (i.e. workers must be no less than 62 years old and have made qualifying contributions for not less than 38 years) (OECD Citation2021).13 In most industrialized countries, the growth of wages in recent decades has been lower than that of labor productivity, resulting in a decline in the share of value added attributable to paid employment (Istat Citation2018). The growth rate of payroll wages has been particularly low in Italy, where average wages declined by around 5% from 2006 to 2015 (Istat Citation2018).14 To ease the comparison with the baseline model, we report fixed effect results in Appendix Table A1. As can be seen, results are qualitatively the same but generally larger in magnitude, consistent with the positive bias detected so far. Interestingly enough, OLS estimates suggest a positive impact on per capita self-employment income, which however is not confirmed by the FHS estimates.15 According to Istat data, the unemployment rate in Southern regions in 2019 was 17.9%, versus 6.6% in the North-West. See http://dati.istat.it.16 The share of the population with tertiary education in 2020 in Italy is 21.3% in the North, 24.2% in the Center, and 16.2% in the South. Data are available at https://italiaindati.com/laureati-in-italia/.17 We report fixed effect estimates of the heterogeneous effects in Appendix Tables A2 and A3. As can be seen, the results are again qualitatively similar and slightly larger in magnitude, thus increasing the confidence in our main results.18 From a geographical perspective, Italy is partitioned into 610 LLS, 107 provinces, and 20 administrative regions.19 The first stage F-test is well below 10.Additional informationFundingWe acknowledge financial support from TIM-LAB (Turin) and Ministero dell'Istruzione, dell'Università e della Ricerca, Award TESUN - 83486178370409, finanziamento dipartimenti di eccellenza, CAP. 1694 TIT. 232 ART. 6.