PDF Archive

Easily share your PDF documents with your contacts, on the Web and Social Networks.

Share a file Manage my documents Convert Recover PDF Search Help Contact

Hara (2014) .pdf

Original filename: Hara (2014).pdf
Title: The impact of firm-provided training on productivity, wages, and transition to regular employment for workers in flexible arrangements
Author: Hiromi Hara

This PDF 1.7 document has been generated by Elsevier / Acrobat Distiller 8.0.0 (Windows), and has been sent on pdf-archive.com on 29/10/2015 at 09:19, from IP address 77.92.x.x. The current document download page has been viewed 2744 times.
File size: 418 KB (24 pages).
Privacy: public file

Download original PDF file

Document preview

J. Japanese Int. Economies 34 (2014) 336–359

Contents lists available at ScienceDirect

Journal of The Japanese and
International Economies
journal homepage: www.elsevier.com/locate/jjie

The impact of firm-provided training on
productivity, wages, and transition to regular
employment for workers in flexible
arrangements q
Hiromi Hara
Department of Social and Family Economy, Faculty of Human Sciences and Design, Japan Women’s University, 2-8-1,
Mejirodai, Bunkyo-ku, Tokyo 112-8681, Japan

a r t i c l e

i n f o

Article history:
Received 29 August 2011
Revised 2 September 2014
Available online 13 October 2014
JEL classification:
Contingent workers
Non-regular workers
Wage increases

a b s t r a c t
Hara, Hiromi—The impact of firm-provided training on productivity, wages, and transition to regular employment for workers in
flexible arrangements
This paper examines the incidence and density of firm-provided
training for workers in flexible work arrangements – i.e., non-regular employees who are working part-time or with fixed-term
employment contracts – and analyzes the effect of this training
on skills, productivity, and wage growth, using a unique survey
of Japanese workers. Among non-regular employees, those who
work on a full-time basis are found to receive a higher density of
both on-the-job training (OJT) and off-the-job training (Off-JT).
Participation in firm-provided training is shown to improve job
skills and productivity, but does not appear to impact the wage
growth of non-regular workers. However, training participation is

I would like to express my appreciation to an anonymous referee and to the editor for their valuable comments. I would also
like to thank the following for their comments: Reiko Kosugi, Masako Kurosawa, Hiroki Sato, Souichi Ohta, Kouichirou Imano,
Yoshihide Sano, Yuzo Yamamoto, Yoshiaki Oomori, Ryuichi Aritoshi, Mitsuo Ishida, Keisuke Nakamura, Michio Nitta, Daniel
Hamermesh, Shuzo Nishimura, Yoshihiro Kaneko, Tadashi Sakai, Wataru Kureishi, Mari Kan, Koichi Ushijima, Masahiro Abe,
Atsushi Sato, and Daiji Kawaguchi. In addition, I would like to thank the participants of the Japan Institute for Labour Policy and
Training (JILPT) Workshop (February 2010), the Labor Research Group of the Institute of Statistical Research (April 2010), the
2010 Annual Meeting of the Japan Industrial Relations and Research Association (June 2010), the 2010 Fall Meeting of the
Japanese Economic Association, the Kansai Labor Research Group (November 2010), and the National Institute of Population
and Social Security Research Workshop (October 2012). For this paper, I received a research grant from the Tokyo Center for
Economic Research, and Grants-in-Aid for Scientific Research (Research Activity Start-up) and (C) (Grant Nos.: 24830086 and
25380371) from Japan Society for the Promotion of Science. Editorial assistance was provided by Janet Abraham. This study uses
the Survey on Working and Learning by JILPT. The data will be available from JILPT. (http://www.jil.go.jp/english/index.html).
E-mail address: harahiromi@fc.jwu.ac.jp

0889-1583/Ó 2014 Elsevier Inc. All rights reserved.

H. Hara / J. Japanese Int. Economies 34 (2014) 336–359


shown to make the transition from non-regular to regular employment in the current occupation more likely, enhancing the probability of future wage increases. J. Japanese Int. Economies 34 (2014)
336–359. Department of Social and Family Economy, Faculty of
Human Sciences and Design, Japan Women’s University, 2-8-1,
Mejirodai, Bunkyo-ku, Tokyo 112-8681, Japan.
Ó 2014 Elsevier Inc. All rights reserved.

1. Introduction
In Japan, as in other OECD countries, the percentage of workers in flexible arrangements has been
growing, and the limited amount of firm-provided training available to such workers has become a
social concern. This paper examines workers in flexible arrangements who have had the opportunity
to participate in firm-provided training, and the impact the training has on productivity, wage growth,
and the transition to regular employment. A unique survey of Japanese workers will be the basis for
this study.
The total number of employees in Japan in 2008 was approximately 51 million, which can be
divided into two categories: 34 million regular employees and 17 million contingent workers, i.e.,
workers in flexible arrangements. Until the 1980s, most Japanese employees were regular employees.
In 1984, immediately after the ‘‘Labor Force Survey’’ began to gather employee data by employment
format, the percentage of contingent workers, excluding officers, was around 15%. Approximately
30 years later in 2011, the percentage of contingent workers more than doubled to 35.2%.
While regular employees (sei-shain) work with full-time permanent employment contracts and are
inside the framework of what is known as ‘‘the Japanese employment system’’, contingent workers are
outside this system, and have fixed-term contracts or work on a part-time basis. Contingent workers
can be further divided into those who are directly hired by their companies and are on the companies’
payrolls, called ‘‘non-regular workers’’, and workers who have employment contracts with temporary
help agencies and are not on the firms’ payrolls, called ‘‘temporary help agency workers’’.1 The analysis
in this paper deals with the former, i.e., non-regular workers, who are an overwhelming 91.9% of contingent workers.
Firm-provided training has played an important role in Japanese economic growth and in the
human resource management of Japanese firms since the 1970s.2 However, according to the ‘‘Basic Survey of Human Resources Development’’ (Ministry of Health, Labour and Welfare, MHLW), participation
in firm-provided training for non-regular workers in Japan is roughly half the level of regular employees,
reaching only 31% even at its highest in 2005.3 Similar results have been reported for other countries. For
example, in the UK, Germany, Spain and other European countries, it has been shown that workers with
fixed-term employment contracts and those working part-time are less likely to participate in workrelated training compared to regular or permanent employees.4
This increase in workers with fewer opportunities for firm-provided training, namely the increase
in non-regular workers, gives rise to two concerns. One is a decrease in the accumulation of human
capital for society as a whole. Second, there is the potential for a further widening of wage differentials. It is known that wages for contingent workers are currently lower than regular employees,
and if skill acquisition is not taking place, their wages could stagnate, causing wage differentials to

They are further divided into ‘‘part-timers’’(pa-ato), ‘‘side job workers’’ (aru-bai-to), ‘‘contract employees’’(keiyaku-shain),
‘‘commissioned workers’’ (syokutaku), and ‘‘other’’.
Hashimoto and Raisian (1989); Mincer and Higuchi (1988).
Kurosawa and Hara (2008, 2009).
Spain: Albert et al. (2005), Germany: Sauermann (2006), UK: Arulampalam and Booth (1998), Booth et al. (2002), OECD: OECD


H. Hara / J. Japanese Int. Economies 34 (2014) 336–359

grow. Indeed, Kambayashi and Kato (2012) shows that in the 1990s, during Japan’s Lost Decade, there
was a significant shift into ‘‘bad jobs’’ – jobs characterized by low wages, poor benefits, job insecurity,
and fewer opportunities for training.
The Japanese government is concerned with the lack of firm-provided training opportunities for
workers in flexible arrangements, and various policies have been introduced to promote training.
The April 2008 revision of the Part Time Employment Act and the introduction of the Job-Card System5
in 2008 are a couple such policies. Although social awareness of the need for firm-provided training is
high and new policies have been introduced, few studies provide comprehensive information on the
extent and effects of training among contingent workers.
According to human capital theories, training is an investment in human capital and should
enhance the individual’s work skills and productivity, enabling the worker to receive a wage in keeping with this enhanced productivity.6 There have been many empirical attempts to directly measure the
effects of accumulating human capital through training, taking into account unobserved heterogeneity
when studying the impact of various forms of training on wage growth. Most of these studies show
the positive effects of training. (Parent (1999), Bartel (1995), Booth (1993)) As for Japan, some recent
studies found that firm-provided training has a positive effect on wage growth for regular employees
as well as for female workers.7 However, as previously mentioned, the impact of firm-provided training
for workers in flexible arrangements has not been examined. This paper attempts to fill this void by
examining the determinants of who receives training among non-regular workers, and what effect the
training has for non-regular workers as compared to regular workers.
The structure of this paper is as follows: Section 2 characterizes non-regular workers in Japan. Section 3 provides a theoretical discussion of training participation by non-regular workers, and examines
possible reasons for differences in the purpose of training for regular and non-regular workers. Section 4 explains the data used in this study and Section 5 introduces a framework for empirical analysis.
Section 6 reports the estimation results of the determinants for firm-provided training and the impact
of training. Section 7 discusses the estimation results, and the final section presents the conclusion.
2. Workers in flexible arrangements in Japan
In Japan, businesses have been in decline, enduring a long stagnation after the economic bubble
burst in the early 1990s. One human resource strategy that emerged in response to these changes
is the active use of flexible staffing.8 This is one of the reasons Japan’s labor market has seen a large
increase in the number of contingent workers, beginning in the 1990s.
As described above, non-regular workers are predominantly workers with part-time or fixed-term
contracts. However, some employers repeatedly renew fixed-term contracts, making it appear as if
some of the fixed-term workers have permanent employment. According to the MHLW (2005),9 over
30% of the firms automatically renew fixed-term contracts, and for part-time and fixed-term workers,
about 20% of the firms renew contracts over 11 times.
Turning to wages, the ‘‘Basic Survey on Wage Structure’’ conducted by MHLW shows that nonregular workers receive less than 80% of regular employees’ wages. In addition, according to a separate
The ‘‘Job-Card system’’ is a policy for people who have difficulty finding new jobs, i.e. job-hopping part-timers (‘‘freeters’’ in
Japanese), women who have finished raising children, and mothers of single parent families. It allows for such workers to be
supported while participating in job training programs and finding new permanent employment. ‘‘Job-Cards’’ record the results of
career counseling, the participation in job training programs, as well as evaluations issued to participants. Achievements from skill
tests or past jobs, and completion certificates from job training programs, may also be entered into ‘‘Job-Cards’’ for use in job
seeking activities. This system allows for obtaining job skills, and for the visual presentation of skills and past work. Through these
means, the way is paved for the Job-Card holders to become employed, or to become regular employees. For details see the
following: http://www.mhlw.go.jp/english/policy/affairs/dl/job_card_eng.pdf.
Becker (1975).
Kurosawa (2001) did an analysis using establishment data from firms in Kitakyushu City, an industrial district in the southern
part of Japan, and showed that informal training raised the wages of workers. Kawaguchi (2006), Toda and Higuchi (2005) analyzed
panel data for Japanese female workers to show that off-the job training raised a worker’s wages. These results can be interpreted
as firm-provided training having raised productivity.
Morishima (2001) reported on why employers use various types of flexible staffing arrangements in Japan.
Ministry of Health, Labour and Wealth, ‘‘2005 General Survey on Fixed-term Employees.’’

H. Hara / J. Japanese Int. Economies 34 (2014) 336–359


survey by MHLW (2006),10 only about 55% of the establishments with part-time workers raised their
wages in the prior year. This is significantly lower than the ratio of establishments giving raises to their
regular workers (72.9%). Furthermore, among the establishments that raised part-time workers’ wages,
regional market wages were a more likely reference for determining pay-raises for part-time workers
than for regular workers. (Regular workers: 4.0%, Part-time workers: 14.1%)
There are also differences in the human resource management (HRM) systems for regular and nonregular workers. According to MHLW (2011), bonuses, various allowances, and fringe benefits apply to
most regular workers, but these apply to very few of the non-regular workers.11 In addition, the proportion of establishments which conduct personnel evaluations of non-regular workers is also lower
than that of regular workers. (Regular workers: 62.7%, Part-time workers: 36.4%) This indicates that a
large proportion of Japanese firms lack a proper HRM system for non-regular workers.

3. Theoretical discussion
In this section, I would like to confirm the theoretical reasons for why firms provide some of their
non-regular workers with training, and how the purpose of training may differ between regular and
non-regular workers. As I explained in Section 2, the main differences between regular and non-regular employees are the length of tenure and working hours, but even among non-regular workers, tenure and working hours differ significantly. I will focus on these points below.
Here, training will be regarded as the sum of two components, one completely general, the other
completely specific, as Becker explained in his book (1975, p. 30). When discussing training theoretically, we usually distinguish between general training and firm-specific training. However, in reality,
training is neither completely general nor completely firm-specific. When one receives training, it is
possible that his/her productivity will not only increase at the company providing the training, but
also at other firms. The greater the firm-specific component, the greater the effect on wages in the firm
providing the training, relative to other firms and vice versa.
First, let me examine the theoretical reasons for providing non-regular employees with training.
Firm-provided training is an investment in human capital, which means that firms make employees
take training only when a return can be expected. Therefore, expected remaining tenure and working
hours long enough to reap a return on investment are factors taken into consideration even for nonregular employees. Among non-regular employees, there are many different work styles; some work
full-time, but others work part-time, and some work with a 1 year contract, while others work with
long tenures, even lifetime employment. This suggests that non-regular employees who have long
expected tenures or who work full-time are more likely to participate in the firm-provided training
than those with fixed term contracts or those who work part-time.
Next, I will explain the differences in the reasons for training regular and non-regular workers. This
can be explained mainly by the differences in the proportions of two components: general versus firmspecific training, and the length of tenure.
The firm-specific component is considered to be a large part of the training provided to a regular
worker. A regular worker is assumed to be a long term worker who will develop his/her career within
a firm, and there seems to be an implicit agreement about this between the firm and regular workers.
Because long term employment is predicted, there is an incentive for both the firm and the worker to
do firm-specific training, and they are willing to share the cost as well as the return. Therefore, the
proportion of firm-specific component of regular workers’ training is considered to be large.
On the contrary, the general component is considered to account for a large proportion of training
for a non-regular worker. A non-regular worker is someone who is not likely to develop his/her career
within a firm. Therefore, he/she does not have an incentive to receive firm-specific training because
he/she cannot use firm-specific skills at another company, and if so, firms are not willing to pay the
firm-specific training cost for non-regular workers. However, a non-regular worker has an incentive
to receive general training because he/she can use general skills in every company, and the firms have

Ministry of Health, Labour and Wealth, ‘‘2006 Survey on Part-time Workers.’’
Ministry of Health, Labour and Wealth, ‘‘2011 Survey on Part-time Workers.’’


H. Hara / J. Japanese Int. Economies 34 (2014) 336–359

an incentive to provide general training if there is a possibility that the trained workers will not quit
after training, as shown by Acemoglu and Pischke (1999a).
Acemoglu and Pischke (1999a) present that if the probability of a worker quitting after training is
less than 1, then the firm has an incentive to do general training, i.e., if some of those trained continue
to work in the company, the company has an incentive to pay for general training. Most non-regular
workers work with fixed-term contracts, but not every trained non-regular worker quits after receiving training. If this is the case, a firm has an incentive to give general training to non-regular workers.
For example, some employers repeatedly renew fixed-term contracts, making it appear as if some of
the fixed-term workers have permanent employment, as I explained in Section 2.
In conclusion, the proportion of general components is considered to be a large part of non-regular
workers’ training, and the firm-specific component is considered to be large for regular workers’ training. If the training for non-regular workers has a large proportion of general components – i.e. skills
common to a particular type of occupation – then it could have a positive impact on converting to
being a regular worker within that specific occupation. I will further examine this point later.

4. Data
4.1. Main characteristics of the analysis sample
In this paper, I have used data from the ‘‘Survey on Working and Learning’’, conducted from October to December 2008.12 This survey was designed in order to examine how Japanese workers were
working, and how skill development was being implemented.
This survey was implemented using the ‘‘area sampling method’’ to ensure representativeness. The
2005 Population Census of Japan was used as a sampling frame, and the target recovery number was
set at 4,000. Then a sample was drawn from the population in two stages13 and the survey was continued until the target numbers for each category in each location were recovered.
In short, the strategy to ensure representativeness consisted of matching the recovery number of
each sex-age category in each location with the ratio in the population census. Then through a complete collection of the surveys, the same distribution for sex and age as the population census was
realized. When using an area sampling method, it is imperative that the target number is recovered.
With this method the response rate is generally not calculated.
However, in the process of recovering the questionnaires, more than the target number were recovered in some locations. In total 4024 questionnaires were recovered – 4000 respondents for targeting
and 24 surplus questionnaires. Unfortunately, I cannot identify which ones are surplus, and because of
this there is a possibility of diminished representativeness. However, 24 is a small proportion of the
total, so even in the worst case scenario, I do not think this is damaging. To clarify this, I will conduct
a test below to confirm whether the key personal characteristics of my sample workers were statistically different from those of the sample universe.
This survey combined self-administered questionnaires and interviews. In the self-administered
questionnaires the standard questions concerned personal attributes, work place attributes, work formats, and working conditions as of September 2008 (hereafter referred to as FY2008). This survey
investigated firm-provided training in detail, including both OJT and Off-JT. I obtained the variables
for participation in firm-provided training, as well as the hours involved in this training for the period
from April 2007 to March 2008 (FY2007). In addition, the self-administered questionnaire asked about

The ‘‘Survey on Work and Learning’’ was designed and carried out by The Japan Institute for Labour Policy and Training under
the direction of Hiromi Hara, Mei Kagawa, Reiko Kosugi, Masako Kurosawa, Hiroki Sato, Yoshihide Sano and Yuzo Yamamoto. This
research project was conducted by request from the Ministry of Health, Labour and Welfare in Japan.
In the first stage, 300 survey locations with a high probability of proportional representation for ages and workers being
surveyed were chosen. Islands, mountains, and isolated locations were excluded. In the second stage 8 categories were created by
age and sex: 25–29 years, 30–34, 35–39 and 40–44 for male and female. Then worker ratios for these 8 categories were calculated
for each location, and the target recovery numbers in each location were set to match worker ratios and to represent the actual sex
and age composition of the sampling frame.

H. Hara / J. Japanese Int. Economies 34 (2014) 336–359


work hours and monthly wages for both March 2007 and September 2008, i.e., for the time
period both before and after the firm-provided training. These questions were designed for
controlling heterogeneity, as I will explain in the following section when discussing the estimation
An interview format was used to learn about work history and life events covering the time from
junior high school graduation to the present. Turnovers, movements between companies, changes in
work format and occupations, as well as marital and family status were asked. In addition, participation in Off-JT from school graduation to FY2008 was also surveyed. This part will be used in the analysis of how training effects the transition from non-regular to regular employment.
The original sample included 2205 observations of workers who were regular employees and 785
observations of non-regular workers in FY2007. At first I restricted the analysis sample to those working in the private sector. Then to analyze the firm-provided training in FY2007, I excluded those not
working at their current company in FY2007.15 The resulting sample size for regular employees was
2042 and 754 for non-regular workers.
The descriptive statistics for the analysis sample of all employees, regular workers and non-regular
workers, are tabulated in Table 1. Here I will confirm only the results for non-regular workers. Over
80% of non-regular workers are female, and their average tenure is 3.9 years – less than half the tenure
of regular workers. Occupations were predominately in service (23.5%), clerical (22.3%), and sales
(22.3%). In terms of industries, wholesale and retail businesses were the highest with 21.0%, followed
by hotel/restaurants (13.6%) and medical/welfare (12.6%).
Here, I will conduct a t-test to check if my sample ensures representativeness for the personal characteristics shown in the 2005 Population Census of Japan. This result is also reported in Table 1.
The t-test shows that the distribution of sex and age of my analysis sample for all employees of private companies, does not diverge from the sample universe in a statistically significant manner. However, the distribution of occupation and industry differ significantly. The ratios of managerial, sales,
and transportation/communication workers in my sample are larger than the sample universe, and
the ratios of workers in electricity/gas/heat/water, finance/insurance, and hotel/restaurants are also
larger. According to the ‘‘Basic Survey on Wage Structure’’ conducted by MLHW – the representative
governmental statistics for Japan – the average wage growth for regular workers in these occupations
and industries are larger than the other occupations or industries for the time period from FY2006 to
FY2008.16 Consequently, these differences may have a positive effect on the estimation result for wage
growth. Therefore I have acknowledged this issue and have explicitly kept it in mind when interpreting
the results.
To better understand the characteristics of the analysis sample, let me confirm several points.
First, in terms of the number of firms a person worked for, 44.4% of regular workers had no
changes, followed by 26.2% with 2 firms, and 14.4% with 3 firms. In other words, 85% of the regular
workers had worked at less than 3 firms. On the other hand, non-regular workers showed the
highest number of people having worked at 3 firms (29.3%), followed by 2 firms (27.3%), and 4
firms (17.0%). This indicates that non-regular workers change firms more frequently than regular
Finally, in terms of the work format at their current firm, a high percentage of all workers have continued to work in the same format. 96% of the regular employees who began work as regular employees have been working continuously as regular employees. Similarly, 96% of the non-regular workers
have been working continuously as non-regular workers.

Recently, some panel surveys have been conducted in Japan. However, they do not ask detailed questions about training, and
do not allow for the type of analysis I am presenting in this paper.
This paper deals with the impact of training within the firm where training was received, and does not deal with the impact of
training before and after a job change to a different firm.
Here I used the data on ‘‘ordinary workers’’ (ippan-rodosya) as regular workers. The wage growth during this period are 1.69
points for electricity/gas/heat/water, 0.70 points for finance/insurance, and 3.04 points for hotel/restaurants. (The average of
these three industries is 0.22 points, and the average of all industries is 1.29 points.) As for occupations, the wage growth for sales
and transportation/communication workers are 4.5 points and 1.8 points respectively, and the average of all occupations is 0.6


H. Hara / J. Japanese Int. Economies 34 (2014) 336–359

Table 1
Descriptive statistics.

Female (=1 if female)
25–29 years old
Professional, Technical
Technician, Production
Transportation, Communication
Agriculture, Forestry, Fishery
Agriculture, Forestry, Fishing, Mining
Electricity, Gas, Heat, Water
Information, Communication
Transportation, Mail
Wholesale, Retail
Finance, Insurance
Real estate, Leasing
Hotel, Restaurants
Education, Learning support
Medical, Welfare
Other services
hOther characteristicsi
Tenure (years)
Union (=1 if unionized)
Full-time (=1 if full-time)
Fixed-term contract (=1 if fixed)
Expected length of tenure (=1 if long)

Survey on work and learning

2005 Population Census
of Japan
Employees (all)

Employees (all)

































N. A.
N. A.

N. A.
N. A.





Data: ‘‘2005 Population Census of Japan,’’ and The Japan Institute for Labour Policy and Training ‘‘Survey on Working and
Notes: 1. The industry categories in this table are the same as those of the 2005 Population Census of Japan, and differ from
other tables below in this paper.
2. ⁄⁄⁄ and ⁄⁄ means that a t-test shows a distribution diverging from the 2005 Population Census of Japan by 1% and 5%
respectively, in a statistically significantly manner.
3. Underlines mean that the numbers given in ‘‘Survey on Working and Learning’’ are larger than those of the 2005 Population
Census of Japan.

4.2. Variables relating to firm-provided training
In this paper I will deal with two forms of firm-provided training. One is done on-site (on-the-job
training or OJT), and involves learning from supervisors and colleagues. The other is done away from
the job site (off-the-job training or Off-JT), and involves training implemented away from the workplace. This includes training in a classroom at the firm but at a location separate from the actual work


H. Hara / J. Japanese Int. Economies 34 (2014) 336–359
Table 2
Descriptive statistics for firm-provided training by employment format.
All employees

Non-regular employees




Work hours

Expected remaining tenure


N: 197

N: 399

N: 603

N: 81

OJT number
(a) Receiving advice
(b) Advising others
(c) Learning by watching
(d) Experiencing other areas
(e) Sharing information







Off-JT (=1 if yes)
Days of Off-JT
hIncluding non-participantsi
hNot including non-participantsi













Data: The Japan Institute for Labour Policy and Training ‘‘Survey on Working and Learning’’. (Same for tables below.)
Note: (a) ‘‘Receiving advice’’ refers to receiving advice from superiors or colleagues. (b) ‘‘Advising others’’ refers to giving advice
or guidance to subordinates or colleagues. (c) ‘‘Learning by watching’’ refers to learning through observing how subordinates
and colleagues do the job. (d) ‘‘Experiencing other areas’’ refers to experiences outside what is directly useful in the current job.
(e) ‘‘Sharing information’’ refers to the sharing of work-related information through meetings and other forums. The OJT
number combines the answers from the 5 questions about OJT training. A ‘‘yes’’ to any category would take 1 for a maximum
total of 5.

First, let me define the variables associated with OJT. In the questionnaires concerning OJT in
FY2007 the following 5 categories were surveyed: ‘‘Received instructions or advice from superiors
or colleagues’’ (receiving advice), ‘‘Gave instructions or advice to subordinates or colleagues’’ (advising
others), ‘‘Learned from watching superiors or colleagues as they worked’’ (learning from watching),
‘‘Experienced work in other areas that helped in the current job’’ (experiencing other areas), ‘‘Learned
information useful for work in meetings’’ (sharing information).17 Those that answered ‘‘frequently’’ or
‘‘occasionally’’ were assigned a value of 1, and other responses were assigned a value of 0. In this paper,
OJT includes all activities that enhance the execution of a job in a workplace.
In addition, to examine the impact of frequent OJT, I constructed a variable that combines the
responses to all 5 questions, called the ‘‘OJT number’’. This number indicates how many of the 5 types
of OJT were received, and indicates the breadth of OJT. These are the 6 variables I will use for OJT in the
following analysis.
Next let me define the Off-JT variable. In this survey, there is a question asking, ‘‘Did your company
send you to a workshop, training session, or somewhere else to acquire knowledge or skills between
April 2007 and March 2008 (FY2007)?’’ The Off-JT variable is 1 if the respondent answers ‘‘yes’’ to this
question, otherwise it is 0. In addition, the number of days of Off-JT will be used to study the training
Table 2 summarizes the descriptive statistics for training variables. Column 1 presents the average
statistics for both samples of regular and non-regular workers, and Column 2 for non-regular workers.
A comparison of Columns 1 and 2 reveals that for all training activities the percentage of all employees
who received training is higher than for non-regular workers, and at the same time it indicates that
regular workers have more opportunities to receive training than non-regular workers.

For each question the respondents were asked to choose one from the following five options, ‘‘frequently’’, ‘‘occasionally’’,
‘‘seldom’’, ‘‘never’’, and ‘‘there was no such person’’.
For those answering ‘‘yes’’ to this question, the survey asked how many days of training the workers received during April
2007–March 2008. The response options were as follows: 1. about 1/2 day, 2. about 1 day, 3. more than 2 days but less than
1 week, 4. more than one week but less than 2 weeks, 5. more than 2 weeks but less than 1 month, and 6. over 1 month. For
answers covering a range, the variable uses the median value. This variable assigns 0 to those who did not participate in Off-JT.


H. Hara / J. Japanese Int. Economies 34 (2014) 336–359

Lastly, to confirm the validity of the data, I compared the participation rate for Off-JT in the ‘‘Survey
on Working and Learning’’ with that of the ‘‘2007 Employment Status Survey (ESS)’’19. The participation rate in Off-JT for workers 25–44 years of age was 33.7% according to the ESS, and when the survey
was broadened to include all workers and recalculated, the Off-JT training percentage was comparable at
36.3%, affirming the information’s validity.

5. Econometric model
The first analysis examines the characteristics of non-regular workers who are sent to firm-provided training. This analysis sample is limited to the non-regular workers.20 Let T i;FY2007 be a dummy
variable which takes 1 if worker i receives training in FY2007, and 0 if no training is received. The conditional expectation of T i;FY2007 is assumed to be as follows:

EðT i;FY2007 jZ i;FY2007 Þ ¼ PðT i;FY2007 ¼ 1jZ i;FY2007 Þ ¼ UðZ i;FY2007 di Þ


where Z i;FY2007 is the individual attributes including work characteristics such as length of expected
tenure, full or part-time work, and external attributes such as age, sex, education, marital status,
the firm’s characteristics, and each prefecture’s unemployment rate in FY2007.
In addition, the OJT number and the number of days of Off-JT participation will be used to study
training density. For answers about the days of Off-JT that cover a range, the central value of each
range was used to give a continuous variable. Furthermore, this variable assigns 0 to those who did
not participate in Off-JT, and approximately 80% of the workers were in this category in FY2007.
The Tobit model will be used to estimate the determinants for the days of Off-JT. The OLS model will
be used to estimate the determinants for the OJT number.
Secondly, I estimated the impact of firm-provided training in FY2007 on the wage growth from
FY2006 to FY2008 by estimating the following equation, using both regular and non-regular workers
as an analysis sample:

ln wi;FY2008 ln wi;FY2006
¼ D ln wi;t


¼ ci T i;FY2007 þ b1 regular i þ X i b2 þ Y i b3 þ regulari X i b4 þ c2 T i;FY2007 regulari þ ui;t
where ln wi,t is the log of hourly wages in year t, regular is the employment status dummy variable that
takes 1 if the respondent is a regular worker in FY2007, and takes 0 if he/she is a non-regular worker.
The dependent variable is the first order of difference for hourly wages between FY2006 and FY2008.
Xi is female and age category dummy by five year segments, and Yi is other characteristics such as education, etc. Both Xi and Yi are information about FY2007, so they are time-invariant. The parameter c2,
the coefficient of the interaction of T i;FY2007 and regular, corresponds to the training effect for regular
workers. ui,t is the error term.
To the extent that this formulation explains the wage growth for individual workers, it controls for
any bias in the firm-provided training variable caused by the self-selection of workers, as long as the
unobservable heterogeneity – such as ability or motivation which affect both training participation
and wage – is unique to individuals and does not vary over time. By taking the first order of difference
as the dependent variable, factors unique to i should be removed, leaving Xi, Yi and T i;FY2007 which do
not vary with time, and allowing me to assume E (ui,t|Xi, Yi, Ti,FY2007) = 0. This allows Eq. (2) to give a
consistent estimator because the OLS estimator is an unbiased estimator under the exogeneity
assumption. Therefore, I estimate the effect of firm-provided training on the wage growth by estimating Eq. (2).
The ‘‘Employment Status Survey’’ is national governmental data, collected by the Ministry of Internal Affairs and
Communications every 5 years. It is considered one of the most accurate surveys for capturing work activities in Japan. While it
surveys the participation rate of Off-JT, it does not survey for OJT. Therefore I compare only Off-JT here.
The expected remaining tenure variable, which is the main explanatory variable and I will explain in Section 6, is only reported
for non-regular workers.

Related documents

hara 2014
the economic effects of a citywide minimum wage
hayes 2007
hidalgo et al 2014
how did japan rebuild itself
why japan factories work

Related keywords