Skip to main content

Broadband metrics and job productivity: a look at county-level data

Abstract

The impact of broadband access and use continues to transform the socioeconomic landscape placing this twenty-first century infrastructure at the center of current policymaking discourse. Past studies have found a relationship between infrastructure investments and economic productivity. Recent broadband-related studies, however, have focused on general availability or adoption, and do not distinguish which specific aspect of the technology is most associated with productivity. Utilizing cross-sectional county-level data from 2017 and spatial econometric models, this research looked into better understanding the impact of multiple broadband indicators on job productivity, including innovative broader measures of digital inclusion. Results indicate that broader metrics focused on adoption or digital distress had a larger positive impact on job productivity in comparison to measures focused on speed or availability. Moreover, these impacts vary across urban and rural settings. Although the relationships identified are not necessarily causal, an alternative matching technique generally supports the results. These findings suggest that the relationship between broadband and economic productivity should be viewed from a larger, more comprehensive socioeconomic perspective. Future research should focus on looking at these effects over time and assess how policies focused on specific broadband characteristics have impacted growth.

This is a preview of subscription content, access via your institution.

Fig. 1

Source: Prepared by authors based on (Anselin et al. 2006)

Notes

  1. The exception to this is the single variable gathered by Microsoft, which was only available for 2018.

  2. Alesina et al. (1999) used the following formula to calculate this index score: Ethnic = 1 − ∑i (Racei)2, where Racei denotes the share of population by race/ethnic categories.

  3. This dataset is known to have issues when conducting spatial analysis (Grubesic and Mack 2016; Grubesic 2011). Further, recent research has shown has shown that the Form 477 data overstates actual broadband availability, with 13 percent of addresses depicted as having service subsequently failing a “self-check” by using the online availability tool of the corresponding providers (Busby and Tanberk 2020). However, it remains the most commonly used and most comprehensive source of broadband data for the USA.

  4. A spreadsheet with county-level data (available by request) was generated from the map online by manually entering the county values. While Microsoft did not share their detailed methodology, some information alludes to this variable being calculated by server logs.

  5. Note that this is the current speed threshold used by the FCC to define broadband.

  6. A queen contiguity weight matrix is used for Fig. 1; alternative distance-based or 5-nearest neighbor specifications give very similar maps and overall Moran’s I measures.

  7. We implement these tests (and resulting spatial models) in GeoDa 1.14.

  8. These variables are: log of population, industrial diversity, percentage of population over 25 with a bachelor’s degree, and the unemployment rate.

  9. We further note that the interpretation of each \(\theta_{1}\) and \(\theta_{2}\) will vary because the broadband measures are scaled differently (i.e., percentages in some instances, indices in others).

  10. These four broadband variables are highlighted in the gray rows of Table 3 and represent the only significant parameters for either metro or nonmetro areas.

  11. The L1 statistic in Table 7 measures the imbalance of the dataset, with higher values implying larger imbalance.

  12. Only the result for 25/3 (pbbnd253) differs between Tables 6 and 7, perhaps because of the different interpretation of the coefficients. Recall that the coefficients in the spatial models are for a 1-unit increase in the percentage of residents with access to 25/3, while in the CEM regressions they are for the “treatment effect” of being in a high-access county.

References

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Roberto Gallardo.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Gallardo, R., Whitacre, B., Kumar, I. et al. Broadband metrics and job productivity: a look at county-level data. Ann Reg Sci 66, 161–184 (2021). https://doi.org/10.1007/s00168-020-01015-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00168-020-01015-0

JEL classification

  • O47
  • R11
  • R12