Content uploaded by Olena Ivus
Author content
All content in this area was uploaded by Olena Ivus on Nov 15, 2019
Content may be subject to copyright.
Canadian Journal of Economics / Revue canadienne d’´
economique, Vol. 48, No. 5
December 2015. Printed in Canada / D ´
ecembre 2015. Imprim´
e au Canada
0008–4085 / 16 / 1803–1830 /©Canadian Economics Association
The employment and wage impact of
broadband deployment in Canada
Olena Ivus and Matthew Boland
Queen’s School of Business, Queen’s University
Abstract. Hundreds of millions of dollars are spent by the Canadian federal and provin-
cial governments to subsidize broadband deployment. This paper provides the first empir-
ical assessment of the impact of broadband on employment and wage growth in Canada.
Variation in elevation explains the regional difference in broadband coverage and is used
as an instrument to estimate the causal effect. We find that the deployment of broadband
in 1997–2011 promoted rural employment and wage growth in service industries. Goods
industries were not impacted. The findings suggest that broadband helps service industry
businesses overcome geographical barriers that have traditionally hampered rural growth.
R´
esum´
e.L’impact du d´
eploiement de la transmission `
a haut d´
ebit sur l’emploi et les salaires.
Les gouvernements provinciaux et f´
ed´
eral ont d´
epens´
e des centaines de millions de dol-
lars pour subventionner le d´
eploiement de la transmission `
a haut d´
ebit. Ce texte donne
une premi`
ere ´
evaluation empirique de l’impact de cet investissement sur la croissance de
l’emploi et des salaires au Canada. La variation dans l’´
el´
evation explique les diff´
erences
r´
egionales dans la couverture des communications `
a haut d´
ebit, et est utilis´
ee comme in-
strument pour estimer l’effet causal. On d´
ecouvre que le d´
eploiement de la transmission
`
a haut d´
ebit entre 1997 et 2011 a engendr´
e une croissance de l’emploi rural et des salaires
ruraux dans les industries de service. Les industries de biens n’ont pas ´
et´
e affect´
ees. Ces
r´
esultats montrent que la communication `
a haut d´
ebit aide les entreprises dans le secteur
des services `
a surmonter les barri`
eres g´
eographiques qui traditionnellement ont nui `
ala
croissance rurale.
JEL classification: L9, O3, R1
1. Introduction
Governments worldwide are spending billions of dollars subsidizing broadband
deployment, and Canada is no exception. Much of these funds are allocated
to rural and remote areas, where commercial broadband deployment is not
The authors would like to thank Daniel Winters of Industry Canada for providing a description
of the broadband data and its sources. The authors would also like to thank Jean-Etienne De
Bettignies and two anonymous reviewers for their helpful comments. Financial support from
SSHRC, The Monieson Centre at Queen’s School of Business and the Eastern Ontario Regional
Network (EORN) is gratefully acknowledged.
Corresponding author: Olena Ivus, oivus@business.queensu.ca
1804 O. Ivus and M. Boland
motivated.1The investments were intended to spur economic activity in these
areas and promote regional growth and development. It is generally understood
that broadband helps to overcome geographical distance by providing individuals
and firms in remote areas with the same opportunities that exist in metropolitan
centres. The evidence suggesting that Internet connectivity lowers the cost of do-
ing business in distant locations supports this conclusion.2And while it has been
over 15 years since the first introduction of broadband, our understanding of
the actual economic impact of broadband availability is limited.3The major un-
resolved question is: How has the deployment of broadband impacted economic
activity and regional growth?
This paper evaluates the impact of broadband deployment on regional em-
ployment and wage growth. Our analysis uses the National Broadband Coverage
data, which provide detailed records of broadband availability across Canada
at various points in time. Our sample covers 4,344 communities representing 76
economic regions (ERs) over the 1997–2011 period. The sample allows for a
comparison between rural and urban regions and spans a sufficiently long period
to allow for the impact of broadband investment to be realized and quantified.
The data’s high level of detail and long time series also allow us to address sev-
eral econometric and data challenges. However, the key empirical challenge is to
credibly identify a causal effect from broadband deployment to economic activity.
We argue that geography provides the necessary source of exogenous variation
in broadband deployment. Specifically, we examine broadband deployment rate
at the level of economic region and use the variation in elevation within each
region as the instrument. The rationale for the instrument is that elevation vari-
ation affects the cost of deploying broadband and so explains the difference in
broadband coverage across regions.
We find that the deployment of broadband in 1997–2011 promoted growth in
aggregate employment and average wages in rural regions across Canada. This
impact is limited to service industries—goods industries are not impacted. The
impact is most pronounced in industries with high intensity of information tech-
nology (IT) use. Rural employment growth in IT-intensive industries declined
when broadband was limited, but rose as broadband became more available. We
also found that while broadband promoted employment growth in service sectors
of rural regions, it curtailed such growth in urban regions. This finding suggests
that broadband helps service industry businesses to overcome geographical bar-
riers that have traditionally hampered rural employment growth, and in so doing,
1 Extending broadband to rural and remote communities has been a goal of the Federal
Government of Canada since 2000. The major early initiative is the Broadband for Rural and
Northern Development program, which was launched in September 2002 as a three-year pilot
program. Over $80 million was invested through this program funding 63 projects for the
implementation of networks to build broadband infrastructure. The more recent major initiative
is Broadband Canada: Connecting Rural Canadians Program, which ran from 2009 until 2012
and directed $225 million into 84 projects.
2 See, for example, Forman et al. (2005a).
3 Broadband first appeared in Canada in 1997 (Czernich et al. 2011).
Impact of broadband deployment in Canada 1805
limits the urban/rural employment gap. Regarding wage growth, the impact of
broadband deployment is the same across rural and urban regions.
To put these findings into perspective, we evaluate the impact under the sce-
nario that all communities within a given economic region moved from having
zero broadband coverage in 1997 to enjoying coverage by any one broadband
technology in 2012. Our estimates predict that in such a scenario, employment
growth in service industries will rise by 1.17 percentage points per year in rural
regions and fall by 1.21 percentage points per year in urban regions, while average
wage growth in service industries will rise by 1.01 and 0.99 percentage points per
year in rural and urban regions, respectively.
Perhaps the biggest challenge in evaluating the economic effects of broadband
deployment is that coverage can be endogenous to economic conditions. Many of
the factors influencing broadband deployment are intricately connected to eco-
nomic activity. Regional population density and income levels, for example, can
impact both the profitability of broadband deployment and regional economic
activity more generally. Further, economic conditions themselves can, directly
and indirectly influence broadband deployment rates. This reverse impact could
result from the Government of Canada’s focus on extending broadband to rural
and remote communities least likely to be served by commercial forces alone.
These communities generally lag behind the others in terms of economic activity.
For these reasons, mere correlation of economic activity and broadband deploy-
ment does not imply causation. In order to identify the true, causal impact of
broadband, it is necessary to isolate exogenous variation in broadband deploy-
ment. We argue that elevation variation within each region can be used for this
purpose.
Elevation variation affects the cost of deploying broadband. It is significantly
cheaper to deploy broadband in areas with limited elevation change. For exam-
ple, microwave wireless systems, such as Multichannel Multipoint Distribution
Service (MMDS), can cover a range of 100 kilometres over flat terrain, but the
coverage range of MMDS is significantly reduced in mountainous areas. Varia-
tion in elevation is also an important consideration for the infrastructure cost of
wired technologies. Corning (2005), for example, notes that the cost of installing
buried wired technologies, such as fiber cable networks, is prohibitively high in
mountainous areas. In the UK, fibre to the cabinet has been dismissed as a viable
option in many regions of Scotland due to challenging terrain.4Our own cor-
respondence with representatives from the Eastern Ontario Regional Network,
which serves to provide high speed internet to residents and businesses in Eastern
Ontario, further confirmed that the cost of installing broadband infrastructure
is increased in areas with varying elevation.
Even if our instrument accounts for significant variation in broadband de-
ployment, the instrument may nonetheless be invalid if it fails the exogeneity
requirement. An important concern in this respect is that elevation variation
could be directly related to economic activity. This relationship could arise, for
4 Source: www.scotnet.co.uk/services/rural-broadband-solutions/bet/
1806 O. Ivus and M. Boland
example, because topography impacts the level of industry agglomeration (Rosen-
thal and Strange 2008). To mitigate this concern, we measure employment and
wages in growth rates, rather than levels. Additionally, we control for factors that
may be related to elevation variation and affect employment or wage growth (i.e.,
population, population density, the degree of urbanization, etc.).
The association between broadband deployment and economic growth has
been studied in several papers (Crandall et al. 2007, Gillett et al. 2007, Shideler
et al. 2007). More recently, the emphasis in the literature has been on estimat-
ing the causal effects. For example, Czernich et al. (2011) estimated the effect of
broadband infrastructure on economic growth in OECD countries in 1996–2007
and found that broadband penetration raised annual per capita growth. Forman
et al. (2012) examined how investment in advanced Internet applications by busi-
ness related to wage and employment growth in US counties between 1995 and
2000. The study found that investment in the Internet contributed to 28% of wage
growth, yet this growth was restricted to only 6% of US counties. These coun-
ties already had relatively high income, high populations and high skills prior to
1995, while the comparative economic performance of isolated and less densely
populated counties did not improve. Kolko (2012) examined economic activity
in the U.S. in 1999–2006 and found that broadband expansion promoted pop-
ulation and employment growth in IT-intensive industries, particularly in areas
with lower population densities, but did not affect average wage and employment
rate.
This paper combines and extends the approaches adopted in the literature,
and as such we owe much to previous work. We use the variation in elevation
within each region as the instrument. This is similar to the approach in Kolko
(2012), where the average slope of the local terrain is used as an instrument for
broadband expansion. The author argues that the cost of extending broadband to
areas with steeper terrain is high.5We also follow Czernich et al. (2011), Forman
et al. (2012) and Kolko (2012) by using data over several years to focus on growth
rates (and not the levels) of economic activity. As did Forman et al. (2012), we
analyse data in long differences. We compare employment and wages in 1997, the
year broadband first appeared in Canada, to those in 2011.
This paper differs from the earlier literature in three important respects. First,
this is the only study to evaluate the impact of broadband deployment on eco-
nomic activity in Canada. Second, since Canada was the first country to introduce
broadband, our data allow analysis of economic growth over longer time peri-
ods. This is important since longer time periods are required to cover the full
adjustment of the economy to broadband deployment. In comparison, the time
periods considered in Forman et al. (2012) and Kolko (2012) are relatively short:
1995–2000 and 1999–2006. Also, Internet infrastructure capabilities in 1995–
2000 were relatively weak compared to the broadband infrastructure deployed in
later years. Third, our study distinguishes between goods and service industries.
5 In addition, Dinkelman (2011) used land gradient as an instrument for project placement when
studying the employment effects of household electrification in rural South Africa.
Impact of broadband deployment in Canada 1807
Such distinction is critical as the entire impact of broadband deployment is real-
ized in service industries, while the aggregate impact is weaker.
The rest of the paper proceeds as follows. Section 2 discusses theoretical foun-
dations. Section 3 outlines our empirical strategy. Section 4 describes the data
on broadband coverage, employment and demographic characteristics and ele-
vation variation. We examine the relationship between broadband deployment
and elevation variation in section 5. The results are presented and discussed in
section 6. Section 7 explores the sensitivity of the results, and section 8 concludes.
2. Theoretical foundations
The theoretical literature provides valuable insights into the complex relationship
between IT adoption and economic outcomes. The Internet is classified as a
General Purpose Technology (Jovanovic and Rousseau 2005, Lipsey et al. 2005),
the adoption and productivity benefits of which vary across regions or industries
(Ristuccia and Solomou 2014).
Several theories highlight rural/urban geographical differences in the benefits
of Internet deployment. Forman et al. (2005a), for example, provides an in-depth
discussion of two contradictory theories, labelled urban leadership and global
village. The urban leadership theory predicts that urban firms face relatively low
costs of Internet adoption because urban regions provide greater access to com-
plementary infrastructure and support resources. As a result, urban firms adopt
the Internet more quickly and receive greater benefits from Internet technology.
The global village theory, in turn, predicts that rural firms face relatively high
marginal returns from Internet adoption, because Internet access reduces com-
munication and coordination costs of doing business in remote areas and helps
overcome barriers to business associated with a distant location and small econ-
omy size. Rural firms thus adopt the Internet more quickly, despite relatively high
adoption costs, and benefit from Internet technology disproportionately more.
Forman et al. (2005b) also discusses the industry composition theory. This
theory maintains that location decisions made by firms prior to the Internet
result in industry clusters; this prior clustering of industries leads to regional
differences in Internet adoption. High concentration of IT-intensive industries
in urban areas, in particular, can explain high urban concentration in Internet
technology adoption and account for a comparatively high benefit of Internet
deployment in urban regions. In a similar vein, the industry composition of cities
is key in determining the impact of improvements in IT in Glaeser and Ponzetto
(2007).
Gaspar and Glaeser (1998) offers an alternative theory that emphasizes the
interaction between electronic and face-to-face communications. Internet tech-
nology may cause some communication to shift from face-to-face to electronic.
This substitution may reduce the benefit of low face-to-face communication costs
offered by urban regions and cause some businesses to reallocate to rural areas.
1808 O. Ivus and M. Boland
At the same time, electronic and face-to-face communication can act as comple-
ments. Internet technology may increase the frequency of electronic communica-
tion, boosting face-to-face communication as well. This complementarity effect
is particularly strong in urban regions, and so urban regions may benefit from
Internet technology relatively more as a result.
The theoretical literature also emphasizes industry differences in the benefits
of Internet deployment. Economic benefits of IT are concentrated in those in-
dustries with high IT investments, given that investment in IT is generally value
producing (Brynjolfsson and Hitt 2000). Pre-Internet investment in IT impacts
Internet investment decisions and determines use of the Internet (Forman 2005).
Forman et al. (2003) further argues that IT innovation will concentrate by in-
dustry and compound over time as long as economic activities are reasonably
stable within firms and across industries. Tambe and Hitt (2013) finds that inter-
corporate movement of IT labour results in productivity spillover across firms.
Those industries in which IT investment is concentrated should see greater
benefits from productivity spillover.
The model in Glaeser and Ponzetto (2007) emphasizes the difference between
idea-producing and goods-producing industries when considering the impact of
improvements in IT. Finance and professional services, for example, belong to
idea-producing industries, while manufacturing is considered a goods-producing
industry. The model predicts that advances in IT may hurt production-oriented
cities but benefit idea-oriented cities.
While the theoretical literature has been useful in identifying a series of links
between Internet deployment and economic outcomes, the direction of the impact
of Internet deployment is not predetermined and becomes an empirical question.
The empirical strategy employed in this paper builds on the above theoretical lit-
erature and examines regional and industry heterogeneity in the economic impact
of broadband deployment.
3. Methodology
To estimate the impact of broadband deployment, we specify the following model:
1Yjt =¯1Bjt +°Xj+®+®t+ejt. (1)
The outcome variable 1Yjt is the employment (or wage) growth in economic
region jover period t. We consider two time periods: t=1, 2. The first period is
from 1997 (the year broadband first appeared in Canada) to 2005 (the first year
of the National Broadband Coverage data). The second period is from 2005 to
2011 (the last year of the Labour Force Survey data). For each period, 1Yjt is
calculated as follows:
1Yjt ≡(log Yj,2005 −log Yj,1997)=(2005 −1997) for t=1,
(log Yj,2011 −log Yj,2005)=(2011 −2005) for t=2. (2)
Impact of broadband deployment in Canada 1809
1Yjt measures the average annual log change in Yjt, which approximates the
average annual percentage change in employment (wage) in region jover period t.
The key independent variable is 1Bjt . It measures the change in broadband
coverage in region jover period tand is defined in section 5. The vector Xj
includes regional controls for initial or permanent characteristics that may af-
fect employment (or wage) growth. Initial controls (for the year 1997) are the
log of population, population density per square kilometre, age distribution (the
percentage of population aged below 15 and the percentage above 65), educa-
tional attainment (the percentage of university and high school graduates) and
firm/establishment size (the percentage of employees employed in small firms,
with less than 20 employees, and the percentage employed in large firms, with
more than 500 employees). The vector Xjalso includes two measures of the
degree of urbanization: the percentage of population living in a census metro-
politan area (CMA)6and an indicator variable for rural economic regions, which
do not contain a CMA.
Our data set is a panel of two time periods. The time series variation allows us
to account for a change in growth over time, which is expected given the shock to
economic conditions brought by the 2008 recession. To do that, we add the time
effect ®t(i.e., the indicator variable for t=2) to (1). Last, ®is a constant and eit
is an error term.
An important concern is that the change in broadband coverage could be
endogenous to employment (or wage) growth. This is very likely for two rea-
sons. First, broadband deployment could be related to a wide range of economic
factors affecting 1Yjt but omitted from (1). Omitting such confounding vari-
ables can create a spurious association between 1Bjt and 1Yjt . Secondly, the
economic conditions themselves can, directly or indirectly, influence broadband
deployment rates, leading to a reverse causality from 1Yjt to 1Bjt. For these rea-
sons, mere association between the economic activity and broadband coverage
does not imply causation. To isolate exogenous variation in broadband, we use
the variation in elevation within region jas the instrument. Elevation variation
affects the cost of deploying broadband and so explains the difference in broad-
band coverage across regions. Our instrumental variable approach is valid under
the key assumption that the variation in elevation within region jdoes not di-
rectly determine j’s employment (or wage) growth. It only affects 1Yjt indirectly
by affecting broadband deployment 1Bjt.
4. Data
4.1. Broadband coverage data
We use the National Broadband Coverage data, compiled by the Canadian Radio-
Television and Telecommunications Commission (CRTC) and Industry Canada.
6 To create this variable, we first identified those economic regions that include a CMA and then
for each economic region, we calculated the percentage of population living in a CMA.
1810 O. Ivus and M. Boland
These data were collected in two separate rounds that differed in scope and detail.
In the first round, the data were gathered at the community level for November
2005. Broadband availability was recorded for 5,426 communities across Canada.
In the second round, detailed coverage maps overlaid with a hexagonal grid were
generated. Industry Canada assigned a unique ID to each hexagoncontaining one
or more Dissemination Block Area (DBA) points.7Broadband availability was
recorded for 49,999 such hexagons,8which correspond to 17,737 different com-
munities across Canada. These data were gathered at several points in time from
July 2009 to March 2012 and were used to evaluate proposals and track progress
for the Broadband Canada Program, which ran during the same period.9Indus-
try Canada solicited feedback from individuals and Internet service providers
regarding the July 2009 data and based on this feedback, revised the data col-
lection process in the following years. We choose March 2011 as the last data
point in our empirical analysis (since 2011 is the last year of our Labour Force
Survey data) and March 2012 as the last data point in our discussion of changes
in broadband coverage over time.
In both rounds, each community/hexagon was polled for the three types of
broadband access technology: Digital Subscriber Line (DSL), Cable Internet
Connection (Cable) and Fixed Wireless Internet Service (Wireless). For each
such technology, data were recorded as a binary variable—taking a value of one
if the technology was available and zero otherwise. A specific type of broadband
access technology is considered available if at least one service provider within
the bounds of a given community/hexagon offers that type of service.
To compile the Broadband Coverage data, the CRTC and Industry Canada
relied upon a number of sources. For wired broadband (i.e., DSL and Cable), the
information on equipment locations, wire centre boundaries and local address
ranges was gathered from the service providers. These data were then used to
estimate coverage areas, either by cross-referencing address ranges and wire centre
boundaries or by measuring coverage radii based on reported hardware capability.
For wireless broadband, coverage areas were estimated using simulated coverage
maps and circular coverage radii around wireless Internet towers.
We examine broadband deployment rate over time and relate it to changes
in economic activity. To begin, we merged the two rounds of broadband data
together. An important consideration in this respect is that the sample of com-
munities differed across the two rounds. The second round was far more compre-
hensive, with a large number of new communities added. These new communities
7 DBA point (or centroid) marks the geographic centre of a Dissemination Block Area, defined by
Statistics Canada as “an area bounded on all sides by roads and/or boundaries of standard
geographic areas. The dissemination block is the smallest geographic area for which population
and dwelling counts are disseminated.” Source: www.statcan.gc.ca/pub/92-195-x/2011001/
geo/db-id/def-eng.htm.
8 Each side of the hexagon is three kilometres long, making the area of each hexagon about
25km2.
9 The data were collected for July 2009, March 2011, November 2011, January 2012 and March
2012.
Impact of broadband deployment in Canada 1811
were relatively underserved and also differed from the rest in terms of geographic
and economic characteristics. These differences between the two samples could
cause endogeneity bias and to preclude this, we limit our data to communities
sampled in both rounds. We used the location information on each hexagon to
extract community names from the 2011 data and then matched communities by
name across the two rounds. The matched data set is a balanced panel of 4,541
communities sampled in both rounds.
Our analysis is at the level of economic region (ER).10 The information on
ERs is not provided in the Broadband Coverage data and so our next step is to
incorporate this information. To do this, we utilized the Geographic Information
System (GIS) software to divide Canada into its 76 ERs using the boundaries
defined by Statistics Canada. We then used the hexagon centroid to assign each
hexagon to a corresponding ER (where applicable). Hexagons not assigned to
any ER were dropped (120 hexagons or 0.24%). Similarly, the communities in the
2005 data were assigned to their respective ERs. To accomplish this, we relied on
the expanded hexagon-level data, where hexagons are linked to both communities
and ERs. All but 197 communities in these data correspond to a single ER, and
we focus our analysis on these communities with one-to-one correspondence. Our
final broadband data set contains 4,344 communities, representing 76 ERs.
Figure 1 plots the average broadband coverage by technology over time. Broad-
band first appeared in Canada in 1997 (Czernich et al. 2011). Until 2005, the de-
ployment of broadband was fastest for DSL, followed by Cable. Fixed Wireless
broadband deployment was slow to start but eventually overtook wired broad-
band. In 2005, the average broadband coverage was 41% for DSL, 20% for Cable
and 11% for Wireless. By 2012, Wireless coverage reached 61%, exceeding both
DSL and Cable coverage, which reached 54% and 34% respectively. What are
the implications of this variation of broadband coverage across technologies? To
answer this question, we must consider the technology itself.
The three technologies differ in network infrastructure. DSL uses copper wire-
pairs of local telephone networks. Not to be mistaken with older dial-up tech-
nologies, DSL utilizes the higher frequency bands on these lines, allowing for a
persistent Internet connection without engaging or interrupting standard tele-
phone service. Cable utilizes existing coaxial cable lines of the local cable televi-
sion network and, like DSL, provides persistent connectivity without affecting
existing cable television service. Fixed wireless Internet service does not depend
on wired connectivity to the end user, but rather provides fixed wireless Internet
access through point-to-point links between networks across distant locations
using microwaves or other radio waves.11
10 Statistics Canada defines an economic region as “a grouping of complete census divisions...
created as a standard geographic unit for analysis of regional economic activity.” Source:
www12.statcan.gc.ca/census-recensement/2011/ref/dict/geo022-eng.cfm.
11 Fixed Wireless service must be distinguished from two other types of wireless service. The first is
mobile wireless service, which utilizes cell towers to allow end-users to connect their
smartphones, tablet PCs and other mobile devices. The second is wireless local area networking,
1812 O. Ivus and M. Boland
0
.2
.4
.6
1997 2005 (November) 2012 (March)
DSL Cable Wireless
FIGURE 1 Average broadband availability by technology
The three technologies also differ in connectivity. A connection’s speed—as
perceived by an end user visiting a website, downloading a file, streaming online
video, etc.—is dictated by latency and bandwidth.12 DSL bandwidth capacity can
range from 128 Kbps to 30 Mbps, depending on the distance between the end
user and the DSL provider’s switching station, and the gauge of the copper wire-
pair connecting the points. Lower-end DSL offerings are excluded from Industry
Canada’s definition of broadband connectivity, according to which broadband
service refers to download speeds of 1.5 Mbps or greater. The most popular
variant of DSL is Asymmetric DSL, which dedicates a disproportionate amount
of the bandwidth available to downloads (downstreaming or incoming data),
at the expense of uploads (upstreaming or outgoing data), to better suit the
needs of the average home and business subscriber. Cable bandwidth capacity is
usually greater than DSL. It is generally no less than 1.5 Mbps and can be as
which is a short-range wireless distribution of an underlying wired network and a feature
commonly available in consumer-grade routers.
12 A connection’s latency concerns the amount of time it takes (i.e., delay) for a network packet to
travel from a source device to a destination device and depends heavily on the processing
capability of the networking routers, switches, firewalls and other hardware along the network
path between the two devices. A given connection’s bandwidth is the maximum throughput on
that network. Data transfer is typically measured in bits (b) transmitted per second, usually in
metric units such as kilobits (i.e., 1,000 bits, abbreviated as Kbps) and megabits (i.e., 1,000,000
bits, abbreviated as Mbps). These measures are not to be confused with those typically used to
describe storage, where capacity is measured in bytes (B) and usually in units that are an
exponent of 2, such kilobytes (i.e., 210 or 1,024 bytes, abbreviated as K or KB) and megabytes
(220 or 1,048,576 bytes, abbreviated as M or MB). For conversion purposes, 1 byte is comprised
of 8 bits.
Impact of broadband deployment in Canada 1813
high as 55 Mbps or even greater. On the high end of this spectrum, transmission
speed is heavily dependent on the quality of the cable modem with which the
end user connects, the quality of the cable network, network load and the degree
of oversubscription in the user’s locality. As with DSL providers, Cable Internet
Service Providers typically offer asymmetrical packages where a greater portion
of bandwidth is dedicated to downloads. Wireless speed is comparable to that of
DSL and Cable and is also frequently offered in asymmetric varieties to end users.
Wireless speed may be affected by line-of-sight and non-line-of-sight propagation
problems that are typical of all radio transmissions.
While the three technologies vary in bandwidth capacities and latency limita-
tions, they all provide the minimum connectivity requirements for the majority of
broadband applications and services. As such, DSL, Cable and Wireless exhibit
a strong degree of substitution. In fact, when measuring broadband coverage to
track the progress of the Broadband Canada Program, Industry Canada’s ap-
proach was to focus on the availability of any broadband service, regardless of
technology. Our analysis is consistent with this approach. We treat the three tech-
nologies as perfect substitutes and measure broadband coverage in a location l
using the following index:
Blt =Dlt +Clt +Wlt
3for l=h,k, (3)
where Dlt,Clt and Wlt is the availability of DSL, Cable and Wireless in community
k(i.e., l=k)orhexagonh(i.e., l=h) at time t. The Broadband index Blt is a simple
average of DSL, Cable and Wireless availability, with equal weights assigned to
each technology.
Before we examine broadband deployment rate at the ER level, it helps provide
a general description of the broadband data. Figure 2 shows the distribution of the
Broadband index across communities in 2005 (on the left) and 2012 (on the right).
The index for 2005 takes on one of four possible values: Bkt ={0, 1=3, 2=3, 1},
since the DSL, Cable and Wireless coverage data are recorded as zero or one for
each community. A particular value taken depends on how many types of access
technologies are available in community k. For example, Bkt =0 if a community
is not covered by any technology and Bkt =1=3 if a community is covered by only
one technology, of any type. For 2012, Bkt is not limited to four values, because
the original data are at the hexagon level. It is measured as the average index
across all hexagons within community k:Bkt =Hk
h=1Bht=Hk, where Hkis the
total number of hexagons within community k.
It is apparent from figure 2 that the distribution for 2005 is largely skewed to
the left. Despite fast deployment of wired Internet in the late 1990s and early
2000s, as many as 47% of communities had zero broadband coverage in 2005.
Across those communities which had broadband, most (35%) had only one type
of technology available. The distribution for 2012 is markedly different. The frac-
tion of communities with zero availability dropped to 10% and the fraction of
communities with one type of technology available dropped to 27%. At the same
1814 O. Ivus and M. Boland
0
.1
.2
.3
.4
.5
0 .33 .67 1 0 .33 .67 1
2005 (November) 2012 (March)
Fraction
Broadband index
Graphs by time
FIGURE 2 The Broadband index across communities
time, the fraction of communities covered by more than one technology rose from
17% in 2005 to 58% in 2012.
4.2. Employment and demographic data
Additional data used in the analysis are from the Labour Force Survey (LFS),
which collects information on different employment and demographic character-
istics of the Canadian workforce. We use annual data for the years 1997, 2005 and
2011 on the following variables: total employment, average hourly wages, popula-
tion, the percentage of population aged below 15 and above 65, the percentage of
university and high school graduates, and the percentage of employees employed
in firms with less than 20 and more than 500 employees. The estimates of em-
ployment and wages are detailed by industry, based on the 2007 North American
Industry Classification System (NAICS). To calculate population density, we use
land area data from the 2006 Census of Population.
The LFS contains information on 69 out of 76 ERs in Canada and so in
the analysis that follows, we focus on these 69 ERs.13 To account for the varying
degree of urbanicity across regions, we distinguish ERs based on their urban/rural
status. Regions containing a CMA are designated as urban, with the balance of
regions designated as rural.14 The two groups are roughly equal in size: 38 ERs
are urban and 31 are rural.
13 The following ERs are excluded from the data: 2490 (Nord-du-Qu´
ebec), 4680 (Northern, MB),
4760 (Northern, SK), 5970 (Nechako, BC), 6010 (Yukon), 6110 (Northwest Territories) and
6210 (Nunavut).
14 Statistics Canada defines a CMA as area with an urban core of 50,000 or more and a total
population of 100,000 or more.
Impact of broadband deployment in Canada 1815
4.3. Elevation variation
Variation in elevation within each region serves as an instrument for the rate of
broadband deployment in that region. We calculate elevation variation as the
standard deviation of the mean elevation across all hexagons within each ER. To
generate mean elevation data at the hexagon level, the following procedure was
employed. First, the latitude and longitude coordinates of each hexagon centroid
(available in the National Broadband Coverage data) were plotted on a map using
GIS software. Second, a buffer region at a radius of 10km was created around
each centroid and for each such buffer region, the elevation data were collected
using a digital elevation model.15 Subsequently, the mean elevation around each
hexagon centroid was computed.
5. Explaining broadband deployment
In this section, we examine changes in broadband coverage across 69 ERs for
which the LFS data are available. First, we aggregate the broadband data to the
level of ER. Depending on the level of detail in the original broadband data,
we define the region-specific Broadband index as follows: Bjt ≡Hj
h=1Bht=Hj
for the hexagon-level data and Bjt ≡Kj
k=1Bkt=Kjfor the community-level data,
where Hjand Kjare the total number of hexagons and communities respectively
within a region j. That is, Bjt measures the average broadband coverage across
hexagons/communities within j. Next, we define the broadband deployment rate
as the average annual log change in j’s index over period t:
1Bjt ≡(ln (1 +Bj,2005)−ln (1 +Bj,1997))=(2005 −1997) for t=1,
(ln (1 +Bj,2011)−ln (1 +Bj,2005))=(2011 −2005) for t=2. (4)
We add one to Bjt before taking logs to avoid undefined values and set the initial
level of broadband to zero: Bj,1997 =0.
The rate 1Bjt approximates the average annual percentage change in broad-
band coverage in jover t. It is measured in percentage, rather than level, changes
to allow for a non-linear (specifically, concave) relationship between employment
(or wage) growth and the level of broadband coverage. This is important since
the impact of broadband deployment is expected to be higher at lower levels of
coverage.
Table 1 shows the results of the first stage regression. In column (1), 1Bjt is
regressed on the log of elevation variation, the demographic covariates and the
time effect. It is apparent that the coefficient on the log of elevation variation
is negative (−.007) and highly statistically significant. The estimate implies that
a one standard deviation (which equals .8681) increase in the log of elevation
variation is associated with .0057 log points (or .57 percentage points) per year
decline in the broadband deployment rate.16 The standardized coefficient on the
15 The average number of data points for each buffer region is greater than 400.
16 The appendix presents the summary statistics.
1816 O. Ivus and M. Boland
TABLE 1
Broadband deployment and elevation variation
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Log of elevation variation −.007*** −.004*** −.005*** −.010*** −.005*** −.009*** −.006*** −.007***
(.001) (.001) (.002) (.002) (.001) (.002) (.001) (.001)
Elevation variation (in 1,000s) −.141***
(.029)
Elevation variation squared .210***
(.056)
Rural indicator .004 .004 .002 .006 .006 .006 .003 .005 .004
(.004) (.004) (.003) (.004) (.007) (.003) (.008) (.003) (.003)
% of population living in a CMA −.000 −.000 .001 .000 −.001 .001*** −.002* .000 .000
(.000) (.000) (.000) (.000) (.001) (.000) (.001) (.000) (.000)
Log of population .005*** .004*** .005*** .004*** .006** .004*** .005* .005*** .005***
(.001) (.001) (.002) (.001) (.003) (.001) (.003) (.001) (.001)
Density per km2−.001 .001 −.001 .000 .000 .007*** −.008** .001 .003
(.002) (.002) (.002) (.003) (.003) (.002) (.004) (.002) (.002)
% of high school graduates .041* .051** .008 .058** .064 .041** .041 .041** .011
(.021) (.021) (.023) (.025) (.043) (.020) (.041) (.020) (.018)
% of university graduates .051 .042 .027 .017 .114 .003 .098 .044 .036
(.040) (.040) (.044) (.043) (.069) (.031) (.084) (.036) (.036)
% of population aged below 15 −.028 −.040 −.179*** −.320*** .354*** −.136** .079 −.044 −.030
(.052) (.054) (.066) (.061) (.105) (.055) (.089) (.051) (.048)
% of population aged above 65 .019 .008 −.044 −.115*** .190*** .002 .037 .017 −.008
(.038) (.038) (.043) (.039) (.064) (.037) (.076) (.036) (.025)
Time effect −.004 −.004 −.041*** −.010*** .046***
(.002) (.002) (.003) (.004) (.006)
Constant .007 −.009 .073*** .073*** −.119*** .027 −.016 .008 .021
(.016) (.015) (.018) (.019) (.028) (.017) (.035) (.015) (.013)
Observations 138 138 138 138 138 69 69 69 69
R2.321 .331 .666 .298 .464 .654 .319 .602 .716
Robust F(log of el. var.) 24.74 10.82 11.01 18.86 17.25 14.26 24.55 59.68
Robust F(el. var.) 22.96
Robust F(el. var. sq.) 14.04
Robust F(el. var., el. var. sq.) 13.55
NOTES: ***, ** and * denote 1%, 5% and 10% significance levels, respectively. Standard errors in parentheses are robust and clustered by ERs.
Impact of broadband deployment in Canada 1817
log of elevation variation is −0.3782, while the standardized coefficient on the
log of population, for example, is −0.0059. The estimate also implies that an
increase in the log of elevation variation from its minimum of 1.2539 (Winnipeg,
MB) to its maximum of 6.3222 (Lower Mainland–Southwest, BC) is associated
with .0333 log points (or 3.39 percentage points) per year decline in the broadband
deployment rate. For comparison, the average change in broadband coverage over
the 1997–2005 period was .0270 log points (2.74%) per year. The results further
indicate that elevation variation is relevant for explaining variation in broadband
deployment; the Fstatistic of 24.74 exceeds its critical value of 10 (Stock et al.
2002).
In column (2), we control for the elevation variation measured in levels, rather
than logs, and also include the square of the elevation variation to allow for a
non-linear relationship between the rate of broadband deployment and elevation
variation.17 It is apparent that our results are not driven by our choice of func-
tional form. The coefficient on elevation variation is negative and the coefficient
on elevation variation squared is positive. Both coefficients are highly statistically
significant. The joint hypotheses test (at the bottom of the table) also indicates
that the two coefficients are jointly statistically significant. These results suggest
that as elevation variation rises, the rate of broadband deployment declines at a
decreasing rate.18
We next analyze the broadband deployment rate by technology. We focus on
DSL in column (3), Cable in column (4) and Wireless in column (5). In column
(3) for example, the rate is defined as in (4), where Bjt ≡Hj
h=1DSLht=Hjfor
l=h,Bjt ≡Kj
k=1DSLkt=Kjfor l=kand DSLlt equals one if location lis covered
by DSL and zero otherwise. We find that the coefficient on the log of elevation
variation is negative and statistically significant in all three columns. Thus, high
elevation variation is associated with lower broadband deployment for all three
technologies. According to the Fstatistic, elevation variation is most relevant for
explaining variation in the deployment of Wireless. This makes sense considering
that elevation variation limits the maximum range of wireless communication
towers, requiring the installation of additional towers at extra cost in mountain-
ous regions. No such additional costs are expected for wired deployments.19
In columns (6) to (8), we re-estimate the regression using cross-sectional, rather
than two-period panel, data. We consider three individual time periods: 1997–
2005 in column (6), 2005–2011 in column (7) and 1997–2011 in column (8). The
coefficient on the log of elevation variation is negative and statistically significant
in all three columns. The Fstatistic is the highest for 1997–2011 and lowest for
2005–2011. One potential reason for the difference in the result is that the period
17 The coefficients in column (2) are rescaled by dividing the level of elevation variation by 1,000.
18 The relationship between the broadband deployment rate and elevation variation is negative for
64 (out of 69) regions.
19 We also considered separately 31 rural and 39 urban regions and found that across regions,
elevation variation explains most differences in broadband coverage in urban areas. Thus,
geography influenced broadband deployment in urban regions more than in rural regions. We
thank anonymous referee for suggesting the analysis.
1818 O. Ivus and M. Boland
of 2005–2011 is relatively short, another is that geography played a larger role in
affecting broadband deployment in the earlier period.
Last in column (9), we consider a different sample of communities: 17,143
communities sampled in the second round, rather than 4,344 communities sam-
pled in both rounds. The analysis utilizes cross-sectional data for the period of
1997–2011. As before, the coefficient on the log of elevation variation is negative
and statistically significant. The Fstatistic rises only when all 17,143 communities
are considered.20
6. Results
6.1. Employment growth
In this section, we estimate the impact of broadband deployment on employment
growth. We first examine aggregate employment and then consider employment
by industry group.
Table 2 reports the aggregate employment growth results. Column 1 shows the
results of our baseline regression (1), where the regressor of interest is the broad-
band deployment rate 1Bjt, as defined in (4). We instrument 1Bjt with the log of
elevation variation variable. The coefficient on 1Bjt measures the average impact
of broadband deployment on employment growth across all ERs. Columns 2 and
3 distinguish ERs based on their rural/urban status. In column 2, we consider
how the relative performance of rural regions is impacted by broadband. The
regressor of interest is the interaction term 1Bjt and the indicator variable for
rural ERs Rj. The instrument here is the interaction between the log of eleva-
tion variation and Rj. In column 3, we also control for the impact of broadband
deployment in urban regions. We include both 1Bjt and 1Bjt ·Rjas regressors,
respectively instrumented by the log of elevation variation and the interaction of
that with Rj. In this specification, the coefficient on 1Bjt measures the average
impact on employment growth across all urban ERs, while the coefficient on
1Bjt ·Rjmeasures the difference in the impact between rural and urban ERs.
It is apparent from column 1 of table 2 that the coefficient on 1Bjt is pos-
itive (.061) but not statistically significant. As such when all ERs are consid-
ered together, the average impact of broadband deployment on the aggregate
employment growth is not statistically different from zero. We next explore if
distinguishing the ERs based on their rural/urban status changes this finding. In
column 2, we replace 1Bjt with 1Bjt ·Rjand find that the coefficient on 1Bjt ·Rj
is positive (.499) and statistically significant at 10% level, while the coefficient
on Rjis negative (−.018) and statistically significant at 5% level. These results
suggest that rural regions lag behind urban ones in terms of the aggregate employ-
20 The broadband deployment rate is measured in log changes in table 1. We show in the online
technical appendix (available in the online version of this article) that the sign of the coefficient
on the log of elevation variation and its significance are not driven by our definition of the
broadband deployment rate. The results remain qualitatively unchanged when the rate is
measured in level or binary changes in tables S1 and S2.
Impact of broadband deployment in Canada 1819
TABLE 2
Aggregate employment growth
Variable (1) (2) (3)
Coeff. St. er. Coeff. St. er. Coeff. St. er.
Broadband deployment rate, 1Bjt .061 .201 −.514* .310
The interaction 1Bjt ·Rj.499* .284 1.024** .458
Rural indicator, Rj−.005 .003 −.018** .008 −.030** .012
% of population living in a CMA .000 .001 .000 .001 .000 .001
Log of population .001 .001 .001 .001 .002 .001
Density per km2−.003 .002 −.002 .002 .002 .003
% of high school graduates −.025 .025 −.023 .025 −.017 .027
% of university graduates .101** .042 .086** .035 .104** .043
% of population aged below 15 .193** .076 .193** .074 .201** .081
% of population aged above 65 −.001 .041 −.020 .036 −.007 .042
% of employees in large firms −.006 .057 .008 .056 .007 .058
% of employees in small firms .018 .065 .032 .064 .022 .069
Time effect −.008*** .002 −.007*** .003 −.008*** .003
Constant −.027 .058 −.032 .055 −.028 .057
First-stage robust F,1Bjt 22.88 16.48
First-stage robust F,1Bjt ·Rj21.29 11.07
Test of endogeneity
robust F0.66 4.74 2.97
p-value .418 .033 .058
R2.228 .093 .028
NOTES: 138 observations. ***, ** and * denote 1%, 5% and 10% significance levels, respectively.
Standard errors are robust and clustered by ERs.
ment growth, but the deployment of broadband works to limit the rural/urban
employment gap. The results shown in column 3, where 1Bjt is also controlled
for, provide a stronger evidence in support of this conclusion. The coefficient
on 1Bjt ·Rjis positive (1.024) and statistically significant at 5% level, while the
coefficient on 1Bjt is negative (−.514) and marginally significant. These results
suggest that the differential impact of broadband deployment in rural regions
is positive; broadband deployment promotes aggregate employment growth in
rural regions more than in urban ones.
Next, we examine employment by industry. In table 3, we consider two dis-
tinct industry groups: goods and services.21 This distinction is critical for our
results. We observe no statistically significant impact on employment growth in
the goods industry group. In the service industry group, by contrast, the signs of
the coefficients on 1Bjt and 1Bjt ·Rjare consistent with those in table 2, and the
statistical significance of the coefficients is noticeably higher. The marginal effect
of broadband deployment, given by @1Yjt=@1Bjt =−.549 +1.082Rj, is positive
(.533) for rural ERs (Rj=1) and negative (−.549) for urban ERs (Rj=0). This
21 Goods industries are: agriculture; resource-based, mining; construction; and manufacturing.
Service industries are: trade; transportation & warehousing; information, culture, recreation;
finance, insurance, real estate; professional, scientific, technical; business, building, other
support; educational services; health care & social assistance; accommodation, food services;
and public administration.
1820 O. Ivus and M. Boland
TABLE 3
Employment growth by industry group
Industry group (1) (2) (3)
Coeff. St. er. Coeff. St. er. Coeff. St. er.
Goods 1Bjt .116 .344 .344 .632
1Bjt ·Rj−.059 .475 −.425 .898
Services 1Bjt .008 .163 −.549** .219
1Bjt ·Rj.533*** .191 1.082*** .301
NOTES: 138 observations. *** and ** denote 1% and 5% significance levels, respectively.
Standard errors are robust and clustered by ERs.
suggests that broadband deployment promotes service employment growth in
rural regions at the expense of urban regions.22
To examine the industry variation more thoroughly, we re-estimate our model
using the data on the industry intensity of IT use documented in Jorgenson et al.
(2012).23 Table 4 shows the results. The outcome variable varies by ERs, industries
and time. In addition to controls in table 2, we include three interaction terms:
between (i) the broadband deployment rate and the industry IT-intensity mea-
sure, 1Bjt ·ITi; (ii) the broadband deployment rate, the rural indicator variable
and the IT-intensity measure, 1Bjt ·Rj·ITi; and (iii) the rural indicator variable
and the IT-intensity measure, Rj·ITi. The instrument for 1Bjt ·ITiis the inter-
action term between the log of elevation variation and ITi, and the instrument for
1Bjt ·Rj·ITiis the interaction term between the log of elevation variation and
Rj·ITi. The specifications in the first two columns also include ITias a separate
control, while the specifications in the last two columns include the set of industry
fixed effects.
From columns 2 and 4 of table 4, the coefficient on 1Bjt ·Rj·ITiis positive
and statistically significant at 5% level, the coefficient on 1Bjt ·ITiis negative and
marginally significant and the coefficient on 1Bjt ·Rjis negative and statistically
insignificant at 10% level. These results confirm that broadband deployment pro-
motes aggregate employment growth in rural regions more than in urban ones
and further indicate that this impact is most pronounced in industries with high
IT intensity. It is also noteworthy that the coefficient on Rj·ITiis negative while
22 The Hausman test of endogeneity rejects the null hypothesis that 1Bjt and 1Bjt ·Rjare
exogenous at 1% level in columns 2 and 3 of table 3.
23 In Jorgenson et al. (2012), NAICS-based industries are classified by their intensity in the
utilization of IT equipment and software. The industries with highest IT intensity include
securities, commodity contracts and investments; professional, scientific and technical services;
management of companies and enterprises; administrative and support services; educational
services; broadcasting and telecommunications; and newspaper, periodical, book publishers.
The industries with lowest IT intensity include: trade; transportation (for all but air
transportation); warehousing and storage; construction; manufacturing; agriculture;
resource-based industries; and mining.
Industries are at 2-digit NAICS level in our paper and 2-, 3- and 4-digit NAICS level in
Jorgenson et al. (2012). To obtain 2-digit IT-intensity measure, we calculate a simple average
across all industries within a given 2-digit industry.
Impact of broadband deployment in Canada 1821
TABLE 4
Employment growth and industry IT intensity
Variable (1) (2) (3) (4)
Broadband deployment rate, 1Bjt −.455* .256 −.495* .180
(.272) (.521) (.260) (.506)
The interaction 1Bjt ·ITi−2.632* −2.507*
(1.389) (1.338)
The interaction 1Bjt ·Rj1.033*** −.308 1.100*** −.046
(.345) (.748) (.338) (.719)
The interaction 1Bjt ·Rj·ITi5.062** 4.320**
(2.336) (2.180)
Rural indicator, Rj−.030*** .007 −.033*** −.000
(.010) (.021) (.010) (.020)
The interaction Rj·ITi−.010 −.150** −.005 −.127**
(.011) (.067) (.010) (.063)
Industry IT share, ITi.027*** .108**
(.004) (.043)
Industry fixed effects included included
First-stage robust F,1Bjt 26.49 15.38 26.55 15.14
First-stage robust F,1Bjt ·ITi12.62 12.57
First-stage robust F,1Bjt ·Rj15.66 11.16 15.5 11.04
First-stage robust F,1Bjt ·Rj·ITi7.61 7.32
Test of endogeneity
robust F5.92 6.57 6.30 6.51
p-value .004 .000 .003 .000
R2.043 .016 .134 .113
NOTES: 1,885 observations. ***, ** and * denote 1%, 5%, and 10% significance levels,
respectively. Other controls included. Standard errors in parenthesis arerobust and clustered
by ERs.
the coefficient on 1Bjt ·Rj·ITiis positive, and both are statistically significant at
5% level. From column 4 specifically, the marginal effect of industry IT intensity
in rural regions, given by @1Yjt=@(Rj·ITi)=−.127+4.3201Bjt , is negative at low
values of the broadband deployment rate, but becomes positive at higher values.24
This result indicates that rural employment growth in IT-intensive industries de-
clines when broadband is non-existent or limited but rises as the availability of
broadband rises.25
6.2. Wage growth
We now use the average hourly wage growth as the outcome variable.26 As before,
we first examine average growth across all industries and then consider growth
by industry group. Table 5 reports the average wage growth results. It is apparent
from the first three rows that broadband deployment promotes wage growth
across all ERs, rural or urban. The coefficient on 1Bjt is larger in column 1
(.368) than in column 2 (.344), suggesting that the contribution of broadband
24 The turnaround value of 1Bjt is 0.0294. The actual value of 1Bjt was above the turnaround
value for 10 (out of 31) rural regions over the 2005–2011 period.
25 We thank anonymous referee for suggesting this comment.
26 When we use average weekly (rather than hourly) wage, the results are very similar.
1822 O. Ivus and M. Boland
TABLE 5
Average wage growth
Variable (1) (2) (3)
Coeff. St. er. Coeff. St. er. Coeff. St. er.
Broadband deployment rate, 1Bjt .368*** .132 .409* .239
The interaction 1Bjt ·Rj.344** .153 −.073 .286
Rural indicator, Rj−.004* .002 −.012*** .004 −.002 .007
% of population living in a CMA −.000 .000 −.000 .000 −.000 .000
Log of population −.002* .001 −.001 .001 −.002* .001
Density per km2−.002 .001 .000 .001 −.002 .002
% of high school graduates −.013 .016 −.010 .016 −.014 .016
% of university graduates .054** .023 .068*** .020 .054** .023
% of population aged below 15 .141*** .043 .147*** .040 .140*** .044
% of population aged above 65 .011 .029 .022 .026 .011 .028
% of employees in large firms .016 .027 .015 .025 .015 .028
% of employees in small firms .080** .033 .071** .030) .079** .033
Time effect .009*** .002 .008*** .001 .009*** .002
Constant −.038 .027 −.035 .025 −.038 .028
First-stage regression robust F,1Bjt 22.88 16.48
First-stage regression robust F,1Bjt ·Rj21.29 11.07
Test of endogeneity
robust F3.77 1.39 1.91
p-value .056 .242 .155
R2.282 .406 .265
NOTES: 138 observations. ***, ** and * denote 1%, 5% and 10% significance levels, respectively.
Standard errors are robust and clustered by ERs.
is .024 points higher in urban regions than rural ones. This difference in impact
however is not statistically significant. From column 3, the coefficient on 1Bjt ·Rj
is not statistically different from zero.
Table 6 reports the results by industry group. In line with the employment
growth results, we find no statistically significant impact on wage growth in the
goods industry group. The results for the service industry group are qualitatively
the same as in table 5, but the coefficients on 1Bjt and 1Bjt ·Rjare higher in
magnitude and the estimates are more precise. Broadband deployment promotes
wage growth in services in both rural and urban regions, with no statistically
significant differential impact.27
6.3. Discussion
In our discussion of results we focus on the service industry groups. From table 3,
the estimates of the impact on rural and urban employment growth are .533 and
−.549, respectively. These estimates imply that a one standard deviation increase
in the broadband deployment rate 1Bjt (which equals .0146 and .0145 in rural
and urban regions, respectively) leads to .0078 log points per year increase in rural
27 The data do not provide evidence that the impact of broadband deployment is more pronounced
in industries with high IT intensity.
Impact of broadband deployment in Canada 1823
TABLE 6
Wage growth by industry group
Industry group (1) (2) (3)
Coeff. St. er. Coeff. St. er. Coeff. St. er.
Goods 1Bjt −.024 .123 .044 .233
1Bjt ·Rj−.080 .150 −.127 .300
Services 1Bjt .459*** .128 .453** .191
1Bjt ·Rj.464*** .159 .011 .232
NOTES: 138 observations. *** and ** denote 1% and 5% significance levels, respectively. Standard
errors are robust and clustered by ERs.
employment growth and .0079 log points per year decline in urban employment
growth. Next, from table 6, the estimates of the impact on rural and urban wage
growth are .464 and .453 respectively. Thus, a one standard deviation increase in
1Bjt leads to a .0068 log points per year increase in rural wage growth and .0066
log points per year increase in urban wage growth.
To put these estimates into perspective, assume for a moment that over the
1997–2012 period, broadband coverage rose from zero (not covered by any tech-
nology) to 1/3 (covered by any one technology) in all communities within a given
economic region. Such change is equivalent to a 0.0204 log points per year in-
crease in 1Bjt. The estimates in table 3 predict that in such a scenario, service
employment growth would rise by 0.0109 log points (or 1.17 percentage points)
per year in rural regions and fall by 0.0112 log points (or 1.21 percentage points)
per year in urban regions. Further, the estimates in table 6 predict that wage
growth in services would rise by 0.0095 and 0.0092 log points (or 1.01 and 0.99
percentage points) per year in rural and urban regions respectively.
Our finding of the positive impact on rural employment growth and a corre-
sponding negative impact on urban employment growth is consistent with the
global village theory discussed in Forman et al. (2005a). The theory predicts
that rural regions benefit disproportionately from broadband deployment be-
cause broadband access reduces the costs of doing business in remote areas and
helps overcome barriers to business associated with a distant location and small
economy size. A similar finding would also arise if businesses relocate to rural re-
gions following broadband deployment because broadband technology reduces
the urban benefit of low in-person communication costs (Gaspar and Glaeser
1998). Our industry-level analysis further shows that the results are driven by
service industries, while goods industries are not impacted. Relative to goods in-
dustries, service industries have relatively high reliance on IT and are also more
IT-skill intensive. Likewise, Kolko (2012) found that positive impact of broad-
band deployment on rural employment growth is strongest in industries that rely
on IT most. Service industries also rely on in-person communication more (in
the absence of the Internet). Additionally, service industries are more footloose,
1824 O. Ivus and M. Boland
meaning they are not tied to any particular location (e.g., because of proximity
to raw materials) and can relocate with little delay in response to a changing
economic environment.
Our results are qualitatively different from Forman et al. (2012), where Inter-
net investment was found to promote wage and employment growth in advanced
urban areas and have no impact elsewhere. One possible explanation for this dif-
ference is that the type of Internet technology studied is different: Forman et al.
(2012) focused on advanced Internet applications while we focus on high-speed
broadband Internet access, which provides improved access to both basic and
advanced Internet services. Generally, benefit from advanced Internet applica-
tions requires highly skilled labour force, which is predominantly concentrated
in urban areas. Another explanation is that Internet infrastructure capabilities
in the 1995–2000 period studied in Forman et al. (2012), were less than those
deployed in subsequent years. In Forman et al. (2012), at most 30% of firms were
using advanced Internet applications in the year 2000. In our study, by contrast,
broadband Internet services are more widespread; the fraction of communities
with zero broadband coverage was 47% in 2005 and 10% in 2012. Furthermore,
all of the firms sampled in Forman et al. (2012) are large (i.e., 100+ employees),
since very few small firms deployed advanced Internet applications at that time.
Our sample, on the other hand, includes firms of all sizes, the vast majority of
which are small.
Our results are qualitatively similar to Kolko (2012), where the focus is on the
expansion of broadband. The magnitudes of the impact are not directly compa-
rable, as the measures of broadband availability used are different: Kolko (2012)
uses the number of broadband providers with subscribers, while we use the num-
ber of access technologies available in a given community.
7. Sensitivity analysis
In this section, we provide further details about the analysis and explore the
sensitivity of our employment growth results to our measure of the broadband
deployment rate, the choice of communities in the sample and the choice of time
period.28 Tables 7 and 8 follow.
We first confirm that our decision to focus on all three technologies together
does not drive our results. To show this, we limit the analysis to Wireless technol-
ogy since from section 5, elevation variation is most relevant for explaining vari-
ation in the deployment of Wireless. The results remain qualitatively unchanged.
From panel 1 of table 7, the coefficient on 1Bjt ·Rjis positive and statistically
significant at 5% level. This result confirms the positive differential impact of
broadband deployment on aggregate employment growth in rural regions. When
we consider two distinct industry groups in table 8, panel 1, we find no statistically
28 The results of the wage growth sensitivity analysis are reported in the online technical appendix,
section S.2.2.
Impact of broadband deployment in Canada 1825
TABLE 7
Aggregate employment growth. Sensitivity
Variable (1) (2) (3)
Coeff. St.er. Coeff. St.er. Coeff. St.er.
Panel 1: Wireless
Broadband deployment rate, 1Bjt .040 .130 −.193 .152
The interaction 1Bjt ·Rj.308* .159 .500** .207
Rural indicator, Rj−.006 .003 −.015** .006 −.021* .008
Panel 2: 1997–2005
Broadband deployment rate, 1Bjt .524 .461 −.321 .565
The interaction 1Bjt ·Rj1.095 .686 1.427 .926
Rural indicator, Rj−.011** .006 −.040* .021 −.048* .025
Panel 3: 1997–2011
Broadband deployment rate, 1Bjt .106 .215 −.502 .314
The interaction 1Bjt ·Rj.564* .307 1.077** .475
Rural indicator, Rj−.006 .003 −.020** .008 −.032*** .012
Panel 4: 17,143 communities
Broadband deployment rate, 1Bjt .096 .195 −.374 .243
The interaction 1Bjt ·Rj.590* .308 .977** .418
Rural indicator, Rj−.006* .003 −.017** .007 −.023*** .008
Panel 5: Fraction control
Broadband deployment rate, 1Bjt .091 .223 −.459 .355
The interaction 1Bjt ·Rj.597** .294 1.074** .490
Rural indicator, Rj−.005 .004 −.019** .007 −.030** .012
Fraction .013 .016 .032* .018 .035* .019
Panel 6: Lagged
Broadband deployment rate, 1Bjt −.335 .430 −1.267 1.089
The interaction 1Bjt ·Rj.265 .464 1.575 1.248
Rural indicator, Rj−.000 .006 −.009 .014 −.040 .031
Panel 7: Placebo test
Broadband deployment rate, 1Bjt −.218 .464 −.064 .589
The interaction 1Bjt ·Rj−.354 .547 −.294 .659
Rural indicator, Rj−.005 .004 .005 .015 .003 .019
NOTES: ***, ** and * denote 1%, 5% and 10% significance levels, respectively. Other controls in-
cluded. Standard errors are robust and clustered by ERs.
significant impact in the goods industry group. In the service industry group, the
coefficients on 1Bjt is negative and statistically significant at 5% level, and the
coefficient on 1Bjt ·Rjis positive and highly statistically significant. Thus, as be-
fore, we find that broadband deployment promotes service employment growth
in rural regions at the expense of urban regions.
We next check if our estimates of the impact are qualitatively unchanged when
we use the cross-sectional data for 1997–2005 and 1997–2011. In table 7, the
coefficient on 1Bjt ·Rjis statistically insignificant in panel 2 and positive and
significant at 5% level in panel 3. Thus, the positive differential impact on ag-
gregate employment growth in rural regions is confirmed for 1997–2011, but not
for 1997–2005. The results by industry groups in table 8, panels 2 and 3, are
1826 O. Ivus and M. Boland
TABLE 8
Employment growth by industry group. Sensitivity
Industry group (1) (2) (3)
Coeff. St.er. Coeff. St.er. Coeff. St.er.
Panel 1: Wireless
Goods 1Bjt .072 .213 .170 .319
1Bjt ·Rj−.037 .297 −.207 .456
Services 1Bjt .005 .112 −.208** .117
1Bjt ·Rj.331*** .104 .529*** .136
Panel 2: 1997–2005
Goods 1Bjt .322 .546 .138 1.070
1Bjt ·Rj.467 .853 .316 1.605
Services 1Bjt .153 .391 −1.088** .541
1Bjt ·Rj1.094** .552 2.208*** .803
Panel 3: 1997–2011
Goods 1Bjt .138 .340 .327 .614
1Bjt ·Rj−.001 .494 −.349 .903
Services 1Bjt .022 .176 −.595*** .223
1Bjt ·Rj.591*** .209 1.187*** .317
Panel 4: 17,143 communities
Goods 1Bjt .118 .285 .228 .414
1Bjt ·Rj−.001 .504 −.241 .739
Services 1Bjt .020 .160 −.474*** .167
1Bjt ·Rj.614*** .195 1.073*** .257
Panel 5: fraction control
Goods 1Bjt .272 .396 .432 .675
1Bjt ·Rj.123 .525 −.357 .981
Services 1Bjt .003 .176 −.505** .219
1Bjt ·Rj.583*** .204 1.091*** .299
Panel 6: lagged
Goods 1Bjt .004 .831 .995 1.756
1Bjt ·Rj−.624 .896 −1.705 2.190
Services 1Bjt −.128 .355 −.889 .733
1Bjt ·Rj.444 .350 1.355* .778
Business, building, support 1Bjt −1.353 1.275 −6.351** 3.117
1Bjt ·Rj4.196** 1.766 11.586*** 4.452
Panel 7: placebo test
Goods 1Bjt −.259 1.077 −.128 1.421
1Bjt ·Rj−.370 1.145 −.248 1.447
Services 1Bjt −.243 .362 −.048 .641
1Bjt ·Rj−.415 .322 −.369 .676
NOTES: ***, ** and * denote 1%, 5% and 10% significance levels, respectively. Other controls in-
cluded. Standard errors are robust and clustered by ERs.
qualitatively unchanged. Broadband deployment promotes service employment
growth in rural regions at the expense of urban regions and has no impact on
employment growth in the goods industry group. The coefficients in panel 2 are
about twice the size of those in panel 3, suggesting that the impact of broadband
deployment on service employment growth is larger over the shorter period.
Impact of broadband deployment in Canada 1827
Thus far, we analyzed 4,344 communities sampled in both rounds, which rep-
resents 80% and 25% of communities sampled in the first and second round
respectively. If communities in the first round were non-randomly sampled, then
our estimates of the impact might be biased. To ensure that the missing data is
not a serious concern, we perform two checks. First, we re-estimate the impact
using the data on 17,143 communities sampled in the second round. This analysis
covers the period from 1997 to 2011. Second, we directly control for the fraction
of communities sampled in both rounds within each ER.29 The results, shown
in panels 4 and 5, confirm our previous findings. Quantitatively, the estimates in
table 8 are comparable to those in table 3. The statistical significance of the esti-
mates rises only when 17,143 communities are considered. The coefficient on the
fraction of communities is only marginally significant in columns 2 and 3, table 7.
In panel 6, we check if the deployment of broadband over the 1997–2005 pe-
riod impacted employment growth over the 2005–2011 period. We do not find
a statistically significant impact of lagged broadband deployment on employ-
ment growth in either the goods or the services industry group. One exception
is Business, building and other support services, which includes three sectors:
(i) Management of companies and enterprises, (ii) Administrative and support
services and (iii) Waste management and remediation services. The results in the
last two rows in panel 6, table 8 show that in this industry, the deployment of
broadband in 1997–2005 promoted employment growth in rural regions at the
expense of urban regions in 2005–2011.
Last, we conduct a placebo test. We check if the deployment of broadband over
1997–2005 impacted employment growth over 1990–1997.30 If our identification
strategy is correct, future broadband deployment should have no effect on past
employment growth. The results in panel 7, tables 7 and 8, pass the placebo test:
none of the key coefficients are statistically different from zero.31
8. Conclusion
This paper studied the impact of broadband deployment on regional employ-
ment and wage growth. Despite the extensive government subsidies for broad-
band deployment, measured in hundreds of millions of dollars in Canada and
billions worldwide, our understanding of the actual economic impact of broad-
band is limited. Perhaps the biggest challenge in evaluating the economic effects
29 The fraction ranges from .11 to .64, with a mean value of .26.
30 Controls included (for 1990): log of population, density per km2, % of high school graduates,
% of university graduates, % of population aged below 15, and % of population aged above 65.
Also included: fraction of communities and constant. The data on other controls at the ER and
CMA level, as well as the data on average hourly wages by ERs and CMAs, are not available for
1990.
31 We also did not find evidence that the deployment of broadband over 1997–2011 impacted
employment growth over 1990–1997.
The broadband deployment rate is measured in log changes in tables 7 and 8. The results for
the rate measured in level or binary changes, shown in the online technical appendix (section
S.2.1), are qualitatively similar.
1828 O. Ivus and M. Boland
of broadband deployment is that coverage can be endogenous to economic con-
ditions. The correlation of broadband deployment and economic growth has
been studied in several papers, but without establishing causation. The empha-
sis of this paper was on estimating the causal effect. The analysis used detailed
records of broadband availability across Canada at various points in time over
the 1997–2011 period. The data’s high level of detail and long time series allowed
us to account for several econometric and data challenges. To credibly identify a
causal effect from broadband deployment to economic activity, the variation in
elevation within each region was used as the instrument.
We found that the deployment of broadband in 1997–2011 promoted growth
in aggregate employment and wages in rural regions across Canada. This impact
was limited to service industries. Goods industries were not impacted. The impact
was most pronounced in industries with high intensity of IT use. Rural employ-
ment growth in IT-intensive industries declined when broadband was limited, but
rose as broadband became more available. We also found that while broadband
promoted employment growth in services in rural regions, it limited such growth
in urban regions. This suggests that broadband helps service industry businesses
overcome geographical barriers that have traditionally hampered rural employ-
ment growth, and in so doing, limits the urban/rural employment gap. At the same
time, rural and urban regions did not differ in the impact on their wage growth.
Our results show that broadband deployment decreases regional disparities in
employment opportunities and enhances the economic viability of rural regions.
In this regard, policies promoting broadband deployment may be said to serve
the objectives of bridging the urban-rural digital divide and reducing regional
inequality. However, our findings also indicate that the observed improvements
in rural employment options come at the expense of those in urban areas. This
second finding leaves open the question of whether increased broadband de-
ployment actually serves the goals of promoting national economic growth and
competitiveness. Judging the true value of broadband deployment to the nation
will require further analyses of the various impacts (e.g., on national productivity,
innovation, business profitability), and these analyses call for future research.
Appendix
TABLE A1
Summary statistics
Variable Obs. Mean Std. dev. Min Max
Aggregate employment growth 138 0.0143 0.0136 −0.0199 0.0557
Urban ERs 76 0.0170 0.0111 −0.0164 0.0403
Rural ERs 62 0.0110 0.0157 −0.0199 0.0557
Aggregate wage growth 138 0.0298 0.0090 0.0107 0.0521
Urban ERs 76 0.0284 0.0075 0.0126 0.0521
Rural ERs 62 0.0316 0.0103 0.011 0.0513
(continued)
Impact of broadband deployment in Canada 1829
TABLE A1
(Continued)
Variable Obs. Mean Std. dev. Min Max
Broadband deployment rate 138 0.0270 0.0151 −0.0249 0.0639
Urban ERs 76 0.0306 0.0145 −.02491 0.0639
Rural ERs 62 0.0226 0.0146 −.01624 0.0545
Log of elevation variation 138 4.4578 0.8681 1.2539 6.3222
Rural indicator 138 0.4493 0.4992 0 1
% of population living in a CMA 138 1.1701 2.5911 0 15.2210
Log of population 138 5.2552 0.9716 3.5056 8.2303
Density per km2138 0.0952 0.3902 0.0002 2.9594
% of high school graduates 138 0.5404 0.0519 0.3842 0.6363
% of university graduates 138 0.1013 0.0384 0.0305 0.2100
% of population aged below 15 138 0.2058 0.0222 0.1622 0.2680
% of population aged above 65 138 0.1183 0.0301 0.0560 0.2094
% of employees in large firms 138 0.2791 0.0749 0.1138 0.4318
% of employees in small firms 138 0.4106 0.0742 0.2736 0.5960
References
Brynjolfsson, E., and L. Hitt (2000) “Beyond computation: Information technology,
organizational transformation and business performance,” The Journal of Economic
Perspectives 14(4), 23–48
Corning, I. (2005) “Broadband technology overview,” Corning Inc. white paper no.
WP6321
Crandall, R., W. Lehr, and R. Litan (2007) “The effects of broadband deployment on
output and employment: A cross-sectional analysis of US data,” Issues in Economic
Policy 6, June
Czernich, N., O. Falck, T. Kretschmer, and L. Woessmann (2011) “Broadband
infrastructure and economic growth,” The Economic Journal 121(552), 505–32
Dinkelman, T. (2011) “The effects of rural electrification on employment: New evidence
from South Africa,” The American Economic Review 101, 3078–108
Forman, C. (2005) “The corporate digital divide: Determinants of Internet adoption,”
Management Science 51(4), 641–54
Forman, C., A. Goldfarb, and S. Greenstein (2003) “Which industries use the Internet?,”
Advances in Applied Microeconomics 12, 47–72
(2005a) “How did location affect adoption of the commercial Internet? Global
village vs. urban leadership,” Journal of Urban Economics 58(3), 389–420
(2005b) “Geographic location and the diffusion of Internet technology,”
Electronic Commerce Research and Applications 4(1), 1–13
(2012) “The Internet and local wages: A puzzle,” The American Economic Review
102(1), 556–75
Gaspar, J., and E. Glaeser (1998) “Information technology and the future of cities,”
Journal of Urban Economics 43(1), 136–56
Gillett, S., W. Lehr, C. Osorio, and M. Sirbu (2006) “Measuring broadband’s economic
impact: Final report,” National Technical Assistance, Training, Research, and
Evaluation Project no. 00-07-13829. Washington, DC: U.S. Department of
Commerce, Economic Development Administration
Glaeser, E., and G. A. Ponzetto (2007) “Did the death of distance hurt Detroit and help
New York?,” NBER working paper no. W13710
1830 O. Ivus and M. Boland
Jorgenson, D., M. Ho, and J. Samuels (2012) “Information technology and US
productivity growth.” In M. Mas and R. Stehrer, eds., Industrial Productivity in
Europe, pp. 34–65. Northampton, MA: Edward Elgar
Jovanovic, B., and P. Rousseau (2005) “General purpose technologies,” Handbook of
Economic Growth 1, 1181–224
Kolko, J. (2012) “Broadband and local growth,” Journal of Urban Economics 71(1),
100–13
Lipsey, R., K. Carlaw, and C. Bekar (2005) Economic Transformations: General Purpose
Technologies and Long Term Economic Growth. Oxford University Press
Ristuccia, C., and S. Solomou (2014) “Can general purpose technology theory explain
economic growth? Electrical power as a case study,” European Review of Economic
History 18(3), 227–47
Shideler, D., N. Badasyan, and L. Taylor (2007) “The economic impact of broadband
deployment in Kentucky,” Regional Economic Development Nov, 88–118
Stock, J., J. Wright, and M. Yogo (2002) “A survey of weak instruments and weak
identification in generalized method of moments,” Journal of Business & Economic
Statistics 20(4), 518–29
Tambe, P., and L. Hitt (2013) “Job hopping, information technology spillovers, and
productivity growth,” Management Science 60(2), 338–55
Supporting information
Additional supporting information can be found in the online version of this
article.
