Main article

View the most recent version.

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.

Archived

This page has been archived on the Web.

1. Introduction
2. Data
3. Measurement of entry and exit
4. Overall patterns of entry and exit
5. Entry and exit, industry dimension
6. Entry and exit, size dimension
7. Conclusion
8. Appendix

1   Introduction

This paper uses Statistics Canada's Longitudinal Employment Analysis Program (LEAP) database to examine firm entry and exit patterns across industries in the Canadian business sector.

The importance of entry and exit is widely recognized. Schumpeterian "creative destruction" models emphasize their role in innovation, and hence, productivity improvement. To survive and to replace incumbents, new firms aggressively adopt new ideas. Pressure from these entrants forces incumbents to be innovative. During this process, winners stay and grow, while losers decline and exit. As well, the product life-cycle model predicts that high turnover (entry and exit) rates are associated with the early stage of life of a new product.

Despite a sizeable theoretical literature, the scarcity of firm-level data restricted empirical analyses of firm dynamics. Since the late 1980s, development of longitudinal micro databases has spurred research around the world, but limitations in the scope and quality of available datasets meant that studies were restricted to specific industries, often manufacturing or retail, or to simple cross-country comparisons (Ahn 2001; Scarpetta et al. 2002; Bartelsman et al. 2009; Baldwin and Lafrance 2011; Baldwin and Gu 2008; Foster et al. 2006; Haskel and Sadun 2009). However, unique features of the LEAP dataset make it possible to derive statistics on firm dynamics for all business sector industries. In addition, a labour-tracking feature in the LEAP dataset allows for merger and acquisition activity to be traced through time, thereby producing more 'organic' rates of entry and exit.

The primary purpose of this report is to provide a descriptive analysis of firm entry and exit patterns in the Canadian economy, and thereby create a solid foundation for future in-depth studies. The 2001 to 2009 vintage files of the LEAP dataset are used to estimate the extent of entry and exit by industry and firm-size for the entire Canadian business sector. In particular, this paper focuses on two aspects of entry and exit. ¹ 

First, the relative importance of entrants and exiters in terms of numbers of firms and employment is outlined. The number of entrants and exiters is a measure of the intensity of entry and exit, since it examines how many individual businesses are involved in this process. Employment in entrants and exiters is a measure of the effect of entry and exit, since it incorporates both intensity and a size dimension. The 'three-year rule' is used to define entry and exit, that is, a firm is deemed an entrant if it appears and lasts one year—a comparison that requires examination of a firm's status across three time periods. The three-year rule distinguishes the numerous short-lived firms that survive for less than one calendar year from more permanent entrants and exiters. Separately identifying these types of firms provides additional information on firm dynamics, and reduces the impact of measurement errors and ill-defined data implicit in these categories of firms. The three-year rule has been applied in several studies of the Organisation for Economic Co-operation and Development (OECD) (Bartelsman et al. 2003).

The persistence of industry entry and exit patterns is also examined over time, and the correlation between industry entry and exit rates is investigated. The results show significant differences in rates across industries and size categories, indicating that industry-specific factors are important in determining entry and exit patterns.

The remainder of this report is organized as follows. Section 2 provides an overview of the LEAP data. Section 3 discusses the measurement of entry and exit using the LEAP database. Section 4 summarizes entry and exit patterns in the total business sector, followed by detailed results by industry in Section 5, and by size, in Section 6. Section 7 concludes.

2   Data

The analysis of firm dynamics requires longitudinal data in order to follow firms through time and identify entries and exits. The Longitudinal Employment Analysis Program (LEAP) dataset makes this possible, in the case of this study, spanning 2000 to 2008. ²  This administrative database includes all firms in the Canadian economy that have some payroll, and therefore, issue at least one Statement of remuneration paid (a T4-slip). LEAP includes incorporated and unincorporated businesses, but excludes self-employed individuals or partnerships where the participants do not draw salaries. Because it is a longitudinal file, the employment level of firms is tracked over time on an annual basis. The data currently cover 1983 to 2008. Based on information gathered by Statistics Canada's Business Register, LEAP data are structured at the level of the "statistical enterprise," which is the lowest level associated with a complete set of financial statements. ³  This statistical unit is referred to as the "firm" in this report.

LEAP's labour-tracking mechanism allows changes in firm structure resulting from merger and acquisition activity (M&A) to be excluded from entry and exit counts. For example, two firms that merge to form a third would not be identified as two exits and one entry in the LEAP file. Rather, the final structure would be preserved, and its employment history would be pushed back through time to maintain consistency. To keep track of these structural changes through time, the dataset at each year is maintained as a different vintage. The last year of each vintage represents the firm structure that existed that year. For this reason, entry and exit rates are calculated based on the last three years of each LEAP vintage. ⁴  This ensures that the most up-to-date information is used in determining birth and death rates, but at the same time, M&A activity is excluded. ⁵  The disadvantage of this method is that it does not enable an analysis of M&A activity in a straight-forward manner, and therefore, such activity is excluded from this study. ⁶ 

LEAP is created using a linkage of the Business Registry along with a summary of employee annual earnings from T4 slips and company payroll remittances. For this reason, the primary variable used to calculate birth and death is the Average Labour Unit (ALU). The ALU is a measure of employment that represents the average employment of an enterprise if it paid its workers the average annual earnings of the typical worker in that industry. ⁷ 

3   Measurement of entry and exit

The literature contains two alternative decision rules for counting firm entry and exit when using an annual dataset. One is based on two-year-period observations. ⁸  Figure 1 presents how firms are categorized by their market appearance under the two-year rule.

Figure 1: The two-year rule of firm counts, by market appearance

A firm with positive employment in year t is considered to be active in that year. An active firm in year t would be counted as an entry in that year if it has no employment record in the previous year, or as a survivor if its employment is positive in the previous year; the firm would be counted as an exit if its employment becomes zero in the next year. Under this rule, the exiting firms in one year are not mutually exclusive from the entering firms or survivors in the same year. As a result, the number of firms by category does not add up to the total number of active firms.

Let the firm counts by category under the two-year rule be the number of active firms (T^I), the number of entrants (E^I) the number of survivors (C^I ), and the number of exiters (X^I ). Thus

An alternative rule for capturing firm entry and exit is based on three-year observations of employment history. ⁹  Figure 2 presents the structure of the three-year rule.

Figure 2: The three-year rule of firm counts, by market appearance

Defining entry and exit over three years instead of two, makes it possible to isolate short-lived firms. A short-lived firm is one that exists for only period t (out, in, out); ¹⁰  an entrant is a firm with positive employment in both periods t and t + 1 (out, in, in); and an exit is defined as having existed in period t and the previous t - 1, but not in t + 1 (in, in, out). Therefore, at any point in time, the population of active firms (T^II ) consists of entrants (E^II), exiters (X^II ), short-lived firms (S^II ), and continuers (C^II ) that show positive employment for all three years observed. Under the three-year rule, all categories are mutually exclusive, and thus, add up to the total number of active firms

The firm counts of entrants, exits, and active firms resulting from the two-year rule and the three-year rules are related. Obviously, the total number of active firms must be the same under both rules. The number of entrants (exiters) under the two-year rule is equal to the number of entrants (exiters) under the three-year rule plus short-lived firms. Also, a survivor under the two-year rule can be either a continuer or an exiter under the three-year rule. Thus

A major advantage of the three-year rule is the additivity of firm counts by market appearance (equation (2)). Consequently, the employment shares of all appearance categories sum to one, which facilitates communication of results. In addition, under the two-year rule, total turnover (the sum of entrants and exiters) is over-stated, because firms entering and exiting the market in the same year are double-counted as both entrants and exiters.

A disadvantage of using the three-year rule with the LEAP dataset is that all measures are referenced in the second-last year in each vintage, and structural change occurring in the last year of the file are not captured. Only entry measures will be affected, as exit measures have the same reference years under both rules. The bias created in the entry rate from the structural change that is not captured under the three-year-rule is assessed by calculating the entry rate referenced to both the last and the second-last years in each vintage (Chart 1). On average, the two series differ very little and track one another over time. Therefore, the three-year rule is used here to calculate entry and exit measures.

Entry and exit measures are calculated using both the number of entrants and exiters, as well as their ALU measure of employment. A firm is considered to be active in year t if its ALU in that year is positive. Entry and exit rates for industry i in year t are calculated using measures of firm counts derived using the three-year rule:

The total entry and exit rates are also calculated in order to compare the results presented here with studies using the two-year rule. These are:

The turnover rate is

which measures the percentage of active firms in a reference year that have undergone a change in their market appearance status in period t. Those short-lived firms are counted only once in this measure. ¹¹ 

Because entering and exiting firms tend to be smaller than continuing firms, it is important to look at their contribution to industry employment. The employment share of Z-category firms for industry i in year t is defined as

Average firm size and its pattern over time provide additional information on firm demographics. The average size of entrants and exiters and their size relative to continuing firms for each industry are calculated as

4   Overall patterns of entry and exit

The target population is the Canadian business sector—all firms excluding public industries and non-profit institutions. In 2008, the number of firms in the business sector employing some labour within the year totaled more than one million.

In any year, four types of firms can be identified: entrants (new firms that did not appear the previous year); exiters (firms that will have exited the market that year); short-lived firms (firms that enter and exit the same year); and continuers (firms that have existed and will continue to exist by year end). Of the total number of firms, continuers are the largest category. Nevertheless, together, entrants and exiters make up 22% to 24% of all firms in any given year. Over the 2000-to-2008 period, firm entry, exit and turnover rates averaged 10.8%, 9.0% and 23.2%, respectively (Table 1).

Although entrants and exiters are numerous, they constitute a small percentage of employment, as measured by average labour units (ALUs). During the 9-year period, firm entry, exit and turnover averaged 1.9%, 1.6% and 3.8% of total employment. Higher intensity (number share) and effectiveness (employment share) of entry than exit at any point indicate vitality and growth of the Canadian economy. The very low shares of employment represented by entrants and exiters, compared with their number shares, are consistent with their small size. Over the 2000-to-2008 period, entrants and exiters averaged 2.1 ALUs (Tables 24 and 25), about one-sixth the average size of firms overall.

Short-lived firms are typically very small, making up about 0.3% of all employment in a given year, and include many self-employed or small venture firms. However, short-lived firms are relatively numerous, accounting for 3% to 4% of all firms and roughly a quarter of entrants and exiters: 23% of entrants were short-lived and exited the same year; 27% of exiters had entered the same year. The difficulty of analyzing these firms is linked to the poor data available for them, including a 25% rate of missing industry classification. As well, inclusion of short-lived firms among both entrants and exiters under the two-year-rule strengthens the correlation between entry and exit.

Entry and exit rates based on the number of firms do not change significantly over time. No clear trend was apparent, with neither rate varying by more than one percentage point over the 2000-to-2008 period (Chart 2). At the aggregate level, the intensity of entry and exit has been relatively stable since 2000.

Unlike firm counts, entry and exit rates weighted by employment show different levels and patterns over time (Chart 4). Entering firms accounted for 2.4% of employment in 2000, but by 2008, the percentage had fallen to 1.5%. The share of employment represented by exiting firms also fell. As a result, turnover in terms of employment dropped steadily throughout the decade. These results reflect the declining size of entering and exiting firms. Over the period, the average size of entrants dropped by 17%, and of exiters, by 30%.

The expected correlation between entry and exit over time is ambiguous, whether based on theory or previous empirical evidence. For a variety of reasons related to market competition and resource reallocation, the "creative destruction" hypothesis and the replacement effect suggest a positive relationship between entry and exit. However, there are other determinants of entry and exit such as business environment and economic growth. Economic growth increases demand, and hence, profits that encourage entry and protect against exit. Empirical evidence in a survey paper by Siegfried and Evans (1994) suggests a lack of consensus about the interaction between entry and exit.

Based on the number of firms, a negative relationship between entry and exit rates emerges at the aggregate level over the 2000-to-2008 period (Chart 2). Distinct periods of increased entry such as 2004 and 2006-2007 coincided with drops in exits. The result is a volatile net entry rate (Chart 3), with clear expansionary periods in 2004 and 2006-2007.

By contrast, because of the simultaneous decrease in the size of entrants and exiters, ¹²  entry and exit rates based on employment were positively correlated (Chart 4). However, their short-run variations were negatively related—again, with troughs in exits in 2004 and 2006-2007. On the other hand, employment from entrants increased slightly in 2003, but then fell. This asymmetric relationship between entry and exit accounted for the sharp increase in the net entry rate of employment during 2003-2004 and the small increase in 2006 (Chart 5).

Overall, in the business sector, the intensity of firm entry and exit is stable, but the average size of entrants and exiters, and hence, their effectiveness in terms of employment share, decreases over time. To reveal inter-industry and inter-size differences in firm entry and exit patterns, the business sector is disaggregated by industry and by firm size.

5   Entry and exit, industry dimension

This section presents the entry and exit measures at 2-digit North American Industry Classification System (NAICS) industries corresponding to private-sector activities. The universe is restricted to private-sector business activities; it excludes firms classified as monetary authorities; primary and secondary schools, universities and colleges; hospitals, offices of physicians, out-patient care centers, ambulatory services, nursing and residential care facilities, and social assistance; private households and religious, grant-making, civic, and professional organizations; and public administration.

Because of delays in business register classification and measurement issues related to accurate firm classification by industry, a substantial number of firms are not assigned a NAICS code early in their existence. For example, for the year 2008, about 24% of entrants, including short-lived firms in the 2009 vintage, have no NAICS code. These unclassified firms are distributed by industry based on the distribution of classified firms.

The descriptive analysis of the industry dimension focuses on three aspects: heterogeneity across industries; the pattern over time; and the inter-industry correlation between entry and exit after correction for fixed, industry effects.

5.1  Heterogeneity across industries

The average measures of entry and exit over the 2000-to-2008 period are reported in Table 2, which includes the average entry rate, exit rate, and the share of the short-lived firms, by both number and employment, and the average size (ALU) of firms in each industry.

Entry

The three entry measures differ considerably across industries. The entry rate based on the number of firms ranged from 6.6% for non-durable manufacturing to 13.5% for professional services. The entry employment share was lowest at 0.7% in utilities and highest at 3.4% in education and art and entertainment. The average size of entrants was lowest at 1.05 ALUs in agriculture and highest at 7.9 ALUs in utilities. Based on number of firms or employment, the service-producing sector had a higher entry rate than did the goods-producing sector, but the average size of entrants in the two sectors was about the same. The two entry rates were positively correlated (0.41); however, both were negatively correlated with the average size of entrants (-0.17 for the rate using number of firms, and -0.55 for the rate using employment).

Exit

The three exit measures also vary across industries. Based on number of firms, the exit rate ranged from 6.0% in health to 11.0% in art and entertainment. Based on employment, the exit rate ranged from 0.6% in utilities to 2.8% in agriculture. The average size of exiters ranged from 0.92 ALU in agriculture to 5.17 ALUs in utility. The two exit rates were higher in the service-producing sector than in the goods-producing sector, but exiters in the two sectors were, on average, almost the same size. The correlation coefficient between the two exit rates was 0.23, smaller than that between the two entry rates. The average size of exiters was negatively correlated with the exit rate calculated using employment (-0.62), but weakly correlated with the exit rate calculated using number of firms (0.04).

Inter-industry relation between entry and exit

At the aggregate level, both entry rates exceeded exit rates during the 2000-to-2008 period. This was generally true at the industry level—whether based on number of firms or employment measures, entry rates surpassed exit rates in all industries except agriculture, mining, and non-durable manufacturing. In agriculture and non-durable manufacturing, both entry rates were lower than the exit rates, thereby contributing to employment contraction in these two industries (Tables 15 and 16). Based on the percentage of firms, the mining industry had more entries than exits; the opposite was true for employment share, reflecting the much larger size of exiters than entrants (Table 2).

Theory predicts that entry and exit are highly correlated across industries. Under the "creative destruction" hypothesis, efficient entrants in an industry may force out less efficient incumbents. As well, the "replacement and resource release" hypothesis (Storey and Jones 1987) suggests that exiters create opportunities for potential entrants. In addition, because of possible connections between barriers to entry and exit, barriers to exit in an industry may discourage entry (Shapiro and Khemani 1987). Empirical evidence in support of the positive inter-industry relation between entry and exit can be found in Shapiro and Khemani (1987), Dunne et al. (1988), Cable and Schwalbach (1991), Dunne and Roberts (1991), and Siegfried and Evans (1992). The results of this paper support these findings. In terms of the industry average over 2000 to 2008, the correlation coefficient was 0.63 between the entry and exit rates calculated using number of firms, 0.87 between the rates calculated using employment, and 0.87 between the average size of entrants and exiters. The positive correlation indicates that an industry with higher-than-average entry rates also tends to have higher-than-average exit rates.

The persistence of industry entry and exit indicates the existence of industry-specific factors behind entry and exit differences. The correlation of entry and exit rates over time is examined to investigate the extent of persistence. A positive inter-temporal correlation indicates that industries with higher-than-average entry (exit) in any one year have higher-than-average entry (exit) levels in subsequent years. Table 3 and Table 4 report the simple inter-temporal correlation of industry entry and exit rates based on the number of firms. Both the entry and exit rates were positively correlated with themselves across different years, and these relationships persisted over time, except for the exit rate in 2000. Exit in 2000 may be largely driven by the dotcom bubble burst. The high persistence of industry entry and exit implies that inter-industry differences are mainly driven by industry-specific factors.

5.2  Patterns over time

Two aspects of industry entry and exit patterns are examined here: time trends and the correlation between entry and exit for each industry.

At the aggregate level, the intensity of entry and exit was stable over time, and the effectiveness of entry and exits decreased, because of declines in relative firm size for both. To determine if these patterns prevailed at the industry level, regressions of entry and exit variables on the time trend variable are performed for each industry (Figure 3).

Figure 3: Regression of firm entry and exit on time trend, by industry, number of firms and employment

Among 18 industries, the entry rate by the number of firms was stable in 9 industries, trended up in 3 industries, and trended down in 6 industries. The exit rate by number of firms was stable in 14 industries and trended down in 4 industries. Entry and exit rates by employment trended down in a majority of industries. ¹³  These industry-level results accord with those derived at the aggregate level.

The correlation between entry and exit over time is calculated in each industry (Table 5). Not surprisingly, the correlation between the entry and exit rates by employment was positive in 16 of 18 industries. This was caused by the decline in the average size of entrants and exiters. The correlation between entry and exit rates by number was negative in 11 industries and positive in 7 industries, implying that entry and exit may react the same way to time-varying factors in some industries, but the opposite in other industries. The positive correlation in the two manufacturing industries used here accords with most empirical findings (Dunne and Roberts 1991; Austin and Rosenbaum 1990; and Siegfried and Evans 1992).

5.3  Inter-industry correlation between entry and exit after correction for fixed industry effects

As discussed earlier, entry and exit rates are generally positively correlated across industries, a relationship that is largely caused by industry-specific factors. Removal of industry averages from entry and exit rates makes it possible to investigate other factors that cause changes over time. Some of these factors may encourage or discourage both entry and exit, while others may encourage one, but discourage the other. If any group of factors dominates over time, consistently positive or negative correlations between entry and exit should be observed. If the same set of factors is not continuously at work, the correlation should alternate from being positive in some periods to being negative in other periods.

Industry fixed effects are removed by de-averaging the industry entry and exit series, and the inter-industry correlations between the entry and exit deviations from the corresponding industry means are calculated over the 2000-to-2008 period.

The inter-industry correlations between entry and exit rates are presented in Table 6 using firm numbers after correcting for fixed industry effects. The row series give the inter-industry correlations between the exit deviations from industry averages in one year and the entry deviations from industry averages in each year from 2000 to 2008. The column series can be interpreted in the same way. No consistent relationship emerged between the entry and exit deviations in the same period in terms of the rates by number of firms. For example, the correlation between entry and exit deviations was negative (-0.42) in 2001 and became positive in 2002 (0.56), which implies that the entry and exit deviations tracked each other across industries in 2002, but moved in opposite directions in 2001.

Because entry and exit may not react to changes immediately, how entry (exit) in one period links to exit (entry) in other periods is also examined. The inter-temporal correlation between the entry (exit) deviations at t and the exit (entry) deviation at varied from being positive to negative when t changes. This indicates that the factors leading to changes over time outside the industry fixed effects vary over time.

However, even if the numbers of entrants and exiters are not always positively correlated, their employment shares should be, because of the 'displacement effect.' To check if this is the case, the temporal and inter-temporal correlations are calculated between the entry and exit deviations when entry and exit are measured by employment (Table 7). The same-period correlations were consistently positive. Such co-movement of the employment shares of entrants and exiters supports the displacement effect.

6   Entry and exit, size dimension

This section disaggregates firm entry and exit by employment. Entrants and exiters are grouped by their ALU measure of employment in the year they enter or exit the market. Because of partial-year market appearance for entrants in their first years and for exiters in their last years, the first-year employment for entrants and the last-year employment for exiters may not represent the size at which their business activities normally function. To address this issue, entrants are also grouped by their second-year employment, and exiters, by their second-last year employment.

Size distribution

The size distribution of entrants based on their first- and second-year ALUs is reported in Table 8. Not surprisingly, entrants were very small. On average, in their first year, 62.2% of entrants had less than one ALU, and 93.2% had fewer than five. The size distribution does not change much in their second year—during the 2000-to-2008 period; less-than-one-ALU firms accounted for 47.7% of total entrants, and less-than-five-ALU firms, 87.7%. Over time, the size distribution of entrants shifted slightly toward smaller firms. Among the 2000 cohort, 63.1% of entrants had less than one ALU in their first year, and 29.3% had one to less than five ALUs. Among the 2008 cohort, the corresponding shares were 64.7% and 30.1%. The shares of all other size categories declined from the 2000 cohort to the 2008 cohort. This pattern persists when based on the second-year size of entrants.

The size distribution of exiters was similar to that of entrants. On average, 65.1% of exiters had less than one ALU in their last year; in their second-last year, the share was 50.4%. An overwhelming majority of exiters had fewer than five ALUs: 93.1% in their last year, and 87.5% in their second-last year. The size distribution of exiters also shifted toward smaller firms. The share of exiters with one to less than five ALUs rose, the share with less than one ALU remained stable, and the share in all other size categories declined (Table 9).

Overall, the size distributions suggest that entrants and exiters are highly concentrated in small firms.

Entry and exit rates by size class

At issue is whether smaller firms are more likely to be new and to be weeded out. Entrants tend to be small relative to continuing firms, indicating a higher share of entrants among small firms. Also, cost disadvantage and scale inefficiency tend to make smaller firms less productive than larger firms, and hence, more likely to fail. Entry and exit rate are calculated by firm size to investigate this issue (Tables 10 and 11).

Whether measured by number of firms or by employment, the entry rate was higher among smaller firms. From 2000 to 2008, the entry rate based on number of firms averaged 19.5% for the smallest size group, 8.5% for firms with one to less than five ALUs, and a mere 1.0% for firms with 100 and more ALUs. The corresponding entry rates based on employment were 17.1%, 7.5% and 0.5%. The lower entry rates by employment than by number of firms suggest that the decrease in entrants' size at the aggregate level is widespread across all size categories. During the period, the entry rate rose only for the smallest size group; the entry rate dropped for all other size groups, particularly the larger ones (Table 10).

The exit rate followed a similar pattern. Smaller firms were more likely than larger firms to exit. The exit rate by number of firms averaged 17.0% for the smallest size group, 6.4% for firms with one to less than five ALUs, and 0.9% for firms with 100 and more ALUs; the employment shares of exiters were 13.6%, 5.7% and 0.5% for the three size categories, respectively. Exit rates based on employment were also lower than exit rates based on number of firms for all size categories. Both exit rates were stable for the two smallest categories and declined for all other size categories over the 2000-to-2008 period (Table 11).

However, in all size categories, more entry than exit occurred.

7   Conclusion

Based on Statistics Canada's Longitudinal Employment Analysis Program (LEAP) dataset, this paper summarizes basic patterns of firm entry and exit in the Canadian business sector, disaggregated by industry and by size dimensions.

Several observations are noteworthy. First, the results consistently show more entry than exit, at the aggregate level and at levels disaggregated by industry and by size. This indicates widespread vitality and growth in the Canadian economy.

Second, the intensity of entry and exit measured by the share of the number firms that are entrants and exiters remains stable over time at the aggregate level and in the majority of industries; meanwhile, the effectiveness of entry and exit measured by employment share decreases over time at the aggregate level and in most industries. The size distributions of entrants and exiters and the entry and exit rates by size class suggest that turnover largely involves small firms, a tendency that has been increasing. As well, the average size of entrants and exiters has fallen over time.

Third, entry and exit rates are negatively correlated over time at the aggregate level; however, at the industry level, these correlations become positive in many industries, including manufacturing and wholesale trade. This implies that time-varying factors affect entry and exit the same way in some industries, but in opposite directions in other industries.

Fourth, industry-specific factors play an important role in determining entry and exit patterns. Not only do entry and exit rates differ considerably across industries, but they persist over time, and the inter-industry correlation between them is strongly positive.

Fifth, after correcting for industry fixed effects, the same time period correlation between industry entry and exit is positive in some years and negative in others. This implies that the impact over time of factors other than industry-specific ones on entry and exit is not consistent. In-depth studies are needed to understand why this is the case and further illustrate the rich analytical capacity of the LEAP database.

8   Appendix