Four Decades of
Creative Destruction: Renewing Canada’s Manufacturing Base
by J. Baldwin and M. Brown*
Market economies are dynamic, with new businesses constantly emerging
and old ones dying. Evidence of this dynamism can be seen almost
every day in the business pages of newspapers or through the casual
observation of the comings and goings of businesses in our local
It has become axiomatic to suggest in public discussions that the degree of dynamic
change is not only large but becoming more so. Jobs are said to no longer exist
for a lifetime. The product lifecycle is described as having shortened. New technologies
are described as coming on line at increasingly faster rates.
Examples of change surround us and act as corroborative evidence for these axioms.
Considerable attention is given to occasions when mature firms fail. Local layoffs
obtain extensive media attention. The high tech boom of the late 1990s and its
subsequent bust is now part of common parlance about how rapidly good times can
turn to bad.
Despite the widespread description of the increasing rapidity of change, and
media coverage of individual events, there is little comprehensive evidence
of the amount of change.1 This
paper remedies this in the area of industrial renewal. It investigates how
long a population of business establishments
takes to renew itself. Does it occur over a decade, or a much longer period?
This paper presents measures of the extent of renewal in Canada’s manufacturing
sector over a four-decade period (1961-1999)—a period that roughly represents
the productive lifetime of a worker. We measure the extent to which plants
that were present in 1961 are renewed forty years later.
The underlying causes of turnover
Renewal occurs when old plants are supplanted by new plants or
when some plants decline and others grow. In both cases, resources
used in production are being shifted from less productive to
more productive plants (Baldwin and Gu, 2002). It is this process
by which new technologies are introduced into plants.
There are many factors that underlie the decisions of firms to downsize or
to close plants. The first is business failure resulting from the inability
of a firm to earn a reasonable rate of return. While failures have unfortunate
consequences for the labour in these establishments, failures also have long-term
benefits. A failure comes from entrepreneurial experimentation and many failed
entrepreneurs learn valuable lessons that enable them to start successful
firms at a later date.
This sort of plant failure is affected by the degree of competition
in the marketplace. Higher levels of domestic or foreign competition
lead to higher levels of plant turnover. Changes in the level of
competition—especially after trade liberalization occurs—may
speed up the renewal process.
Plants may also downsize or close because of changes in technology
or tastes that favour the production of some industries or plants
over others. Technological change may make some plants obsolete,
resulting in the need to build new ones to accommodate the new
technology. Here, renewal is critical to the introduction of new
advanced technologies that create a competitive edge for a country.
The passing of time can itself lead to the closure of plants
as they deteriorate with the ageing process. At some point, the
firm is faced with a decision to reinvest in the plant or to build
a new one. In many instances, it may be more profitable to build
a new plant.
Finally, there may be geographic shifts in demand or sources of important inputs
that change the optimal location of production and lead to plant closedowns
in one location and openings elsewhere. These shifts can occur over relatively
short distances (i.e., when manufacturing firms move from core urban areas
to the suburban fringes of cities—see Brown and Baldwin 2003) or across
provinces, which in Canada’s case has involved a general east to west
movement in economic activity.
In summary, there are numerous forces at work that may cause
an economy to renew its production base. In this paper, we investigate
the size of all of these forces taken together.
Measuring turnover: Renewal and fading rates
Using a database developed from the Census of Manufacturers, what
is now the Annual Survey of Manufacturers (ASM), allows us to follow
individual plants over time. As such, we can determine the employment
of a plant in any given year, when it was born and when it was
shut down, if ever.
Other studies have also examined a phenomenon called job turnover
(see Baldwin and Gorecki, 1990; Davis, Haltiwanger, and Schuh,
1996). Most of these studies are short term in nature. They focus
on one-year or five-year periods and calculate rates of job growth
or job decline—defined as the growth in jobs in plants where
employment is growing divided by initial employment or job loss
divided by initial employment. Studies of job turnover conducted
over short periods may contain considerable noise, as firms may
experience random increases one period only to see these reversed
the next (Davis, Haltiwanger, and Schuh, 1993). These studies cannot
determine whether change is an ephemeral phenomenon or one that
has inexorable cumulative consequences. In this study, we focus
on a much longer period, in order to remove short-run movements
that may be quickly reversed.
In order to develop measures of long-run cumulative change, we measure turnover
over periods of one, two, three and four decades. For each time frame, we
ask what proportion of jobs in the first year was lost by the last, either
because of plant closures or downsizing. We also ask what proportion of jobs
in the last year was new, either because of investments in new plants or
the expansion of incumbent plants. The former is a measure of the extent
to which jobs are fading away, at the level of individual business units;
the latter is a measure of the extent to which the economy is being renewed,
again at the level of individual business units.
In more concrete terms, we describe the renewal rate for the period from 1961
to 1999 as the sum of jobs in 1999 plants that were new since 1961 and present
in 1999 (New Plants) plus the sum of job gains in continuing incumbent plants
whose employment grew between 1961 and 1999 (Job Gains) divided by total
employment in 1999 (Employment1999) or more formally:
The fading rate is similarly defined as the sum of jobs in 1961
plants that closed between 1961 and 1999 (Closed Plants) plus the
jobs that have been lost in continuing incumbent plants that downsized
(Job Losses) over this period divided by employment in 1961 (Employment1961):
Renewal and fading rates can be calculated for any combination
of years between 1961 and 1999: we calculate these rates for time
periods spanning ten, twenty, thirty, and (almost) forty years.2
Renewal and fading rates across time
How long does it take for the manufacturing sector of the Canadian
economy to renew itself? The renewal and fading rates for four
different periods of time are presented in Figure 1. Using decadal
time periods, on average around 40% of jobs result from renewal.
A similar proportion of jobs are lost over a decade. Therefore,
over a relatively short period of time, about two in every five
jobs are eliminated, either because of plant closures or downsizing
and, in turn, are renewed.
Increasing the time frame results, not surprisingly, in a larger proportion
of jobs that are renewed. For instance, using
a 20-year time frame, over 65% of the economy is renewed; over
a 30-year period, renewal amounts to just over 75%; and by 40 years, just over
85% of jobs are new. In 40 years, Canada’s manufacturing economy has
effectively turned over. It is almost completely renewed.
Incumbent job gains and losses versus plant entry and exit?
Turnover and renewal can come from different sources. On the one
hand, it may come from the redistribution of jobs within incumbent
plants—from one group of incumbents who are in decline
to another who are growing. Or it may come from the closing down
of old plants and the opening of new ones. The former provides
a more stable environment since incumbents remain; they just
exchange relative positions. The latter involves a more fundamental
transformation since some participants disappear while other
new ones are created.
Renewal and fading rates are divided into these components in Table 1. The
majority of new jobs are in plants that have been established since 1961. That
is, of the 86% of jobs that were new since 1961,
76 percentage points are due to the establishment of new plants. Of the 81%
of jobs lost since 1961,
72 percentage points were due to plant closures. Therefore, in the long-run,
the renewal process is not primarily the result of shifting workers from declining
to growing plants. Rather, it results from the nearly complete renewal of the
country’s stock of manufacturing plants and jobs.
Table 1. Renewal and fading rates from 1961-1999
|Job losses 
|Job gains 
|Closed plants 
|New plants 
|Growth rate 1961-1999
Our interpretation of the economic effects of these high levels of plant entry
and exit depends on whether we view them as independent of, or dependent
on, each other. If entry and exit are independent of each other, then the
incremental effects of entry and exit that we observe in our daily lives
have a large long-run cumulative effect. Without constant investment in new
plants to renew the manufacturing base of a country, region or city, their
manufacturing sector would almost completely fade away after a generation.
However, if the plant entry and exit depend on each other—entry causes
some exit—then we have to modify our interpretation. Without entry,
many of the plants that exit would probably continue to exist. But because
entrants are more productive than exits, without entry the population of
plants would use older, less productive technology and management practices.
In all likelihood, plant exits result from both the competitive push of rivals
and the loss of plants due to ‘natural causes’—it results
from the twin processes of selection and entropy. From an economic perspective,
therefore, renewal is necessary both for replacing the constant erosion of
the economic base of an economy and for raising its productivity.
Renewal and fading rates across provinces
The high fading and renewal rates experienced nationally between
1961 and 1999 suggest there is considerable potential for geographic
shifts in employment across the country. This would be manifest
by extremely different renewal rates across the country. At issue
is whether we find very high renewal rates in those provinces where
total manufacturing employment grew and much higher fading rates
in those provinces where it fell.
The renewal and fading rates across provinces are plotted in
Figure 2. Provincial fading rates are remarkably similar. From
Nova Scotia west, they range from just under 80% to just over 80%.
The two exceptions to the provincial pattern are Newfoundland and
Prince Edward Island, which both have relatively low fading rates.
These two provinces have by far the smallest manufacturing economies
of all the provinces. So their low fading rates may result from
It is also apparent from Figure 2 that none of provinces experienced a particularly
high exodus of manufacturing employment. That is, no one province stood out
as a location where manufacturing firms were prematurely downsizing or closing
their manufacturing activities.
Renewal rates show a similarly consistent pattern. Only Newfoundland had a
renewal rate that was below 80%. The highest renewal rate was experienced by
Alberta with a rate of 93%. In other words, just over ninety percent of jobs
in Alberta in 1999 are in plants that did not exist in or are new jobs added
to incumbent plants since 1961.
Fading and renewal rates across industries
Since renewal and fading rates vary across provinces, they may
also vary across industries. Variations in renewal rates across
industries occur because of market structure, changes in technology
and differences in underlying possibilities for entrepreneurial
The renewal and fading rates are presented by industry in Table 2 and Figure
3. Industries such as Primary Metals and Transportation Equipment have the
lowest fading rates (in the mid 50’s). These are also industries where
economies of scale and concentrated market structures exist. Refined petroleum,
another highly concentrated industry, had a relatively low fading rate of 73%.
At the other end of the spectrum are rubber, plastics, wood, furniture, and
fabricated metals whose fading rates are in the high 80s. But most industries
fall within a range of about 80% to 90%. The renewal rates also fall within
a band stretching from the low 80s to the mid 90s. The major exception is the
tobacco industry with only a 30% renewal rate.
Table 2. Renewal and fading rates by industry, 1961-1999
||Employment growth rate
|Food and beverages
|Rubber and plastics
|Textile and clothing
Growth should influence both the rate of fading and the rate of renewal3. The
correlation between growth and fading at the industry level is negative as
expected, but quite small (-0.06). On the other hand, growth is positively
corelated (0.54) to renewal. Economies that are growing are renewing themselves.
Economies that are not growing have a slightly higher rate of fading, but
not much. This corresponds to the results reported in Baldwin and Gorecki
(1990) that suggest exit is a more or less random process that occurs with
about the same rate across different industries, geographical areas and time
periods. It stems from the more or less constant rate of probability that
failure will occur. It can respond to economic events, but it does so weakly—at
least compared to entry. Entry and renewal, on the other hand, are more sensitive
to economic conditions. Where there is more growth, there is more entry and
more renewal. Regardless of whether the direction of causation goes from
renewal to growth or from growth to renewal, renewal is associated with growth.
And that is the reason for the widespread interest in the amount of renewal
Previous studies have outlined the existence of high turnover in
the job market. But they did not indicate whether this phenomenon
was the result of stops and starts that reversed themselves, leaving
most producers in the same relative position, or whether this was
part of an inexorable long-term process of growth and decline.
This paper indicates that job turnover in manufacturing has a substantial long-term
component. Moreover, it quantifies the size of this process decade by decade.
Over a decade, almost 40% of jobs are renewed. Using a 20-year time frame,
over 65% of the economy is renewed; over a 30 year period, renewal amounts
to just over 75%; and by 40 years, just over 85% of jobs are new.
It is also the case that this process occurs at similarly high
levels both across regions and industries. There are exceptions
but these are aberrations. This is a phenomenon that offers more
similarities across jurisdictions than differences, as it is pervasive
across the industrial system.
The rates of renewal are closely related to growth. Growth may create renewal
or renewal may lead to growth. But there is no doubt that the two are found
together. Interestingly, rates of fading are less related to growth rates.
Producers go out of business at a rate that is relatively similar across industries;
but renewal is much higher where growth is higher. This explains the widespread
interest in knowing whether a region or an industry is renewing itself.
Finally, most renewal results from entering plants replacing those that have
exited. Entry renews the manufacturing sector in two fundamental ways. First,
without entry, the high rate of exit would lead to a substantial diminution
of the sector. Second, entry leads to the replacement of old, less productive
plants with new, more productive ones. Renewal is necessary for replacing the
constant erosion of the base of an economy and for raising its productivity.
Baldwin, J.R. 1995. The Dynamics of Industrial Competition:
A North American Perspective. Cambridge: Cambridge University Press.
Baldwin, J.R. and P. Gorecki. 1990. Structural Change and
the Adjustment Process: Perspectives on Firm Growth and Worker
Ottawa: Statistics Canada and the Economic Council of Canada.
Baldwin, J.R. and W. Gu. 2003. “Plant Turnover and Productivity
Growth in Canadian Manufacturing.” Analytical Studies Research
Paper Series no. 193. Catalogue no. 11F0019-MIE. Ottawa: Statistics
Brown, W.M. 2004. “Renewing Canada’s manufacturing
economy: A regional comparison, 1973-1996.” Economic Analysis
Research Paper Series no. 023. Catalogue no. 11F0027-MIE. Ottawa:
Statistics Canada. Forthcoming.
Brown, W.M. and J.R. Baldwin. 2003. “The changing geography
of the Canadian manufacturing sector in metropolitan and rural
regions, 1976-1997”. The Canadian Geographer, 47(2): 116-134.
Davis, S., J. Haltiwanger and S. Schuh. 1996. Job Creation
and Destruction. Cambridge, Mass.: MIT Press.
Davis, S., J. Haltiwanger and S. Schuh. 1993. “Small Business
and Job Creation: Dissecting the Myth and Reassessing the Facts”.
NBER working Paper No. w4492. Cambridge, MA: National Bureau of
Recent feature articles
* Micro-Economic Analysis Division (613)951-4676.
1 See Baldwin (1995) for a study that examines
change in the manufacturing sector in the 1970s.
2 For the decadal estimates, we average 1961-69,
1970-79, 1980-89, and 1990-99; for the twenty year period, we use 1961-79, 1970-1989,
and 1980-99; for the thirty year period, 1961-89 and 1970-99 ; and for the forty-year
period, we use 1961-99.
3 See Appendix A in Brown (2004) for a discussion
of the relationship between growth and renewal rates.