Do older workers really reduce firm productivity?

Abstract

In this article, we examine the effect of workforce ageing on company productivity, using an analysis based on Korean firms. We found that an increase in the ratio of workers aged over 50 years to total workers had a negative effect on value added per worker, which was consistent with the findings of most previous studies based on European data. However, the results of the analysis, including various classifications such as size, industry and several financial conditions, revealed that an increase in the ratio of older workers had positive effects on value added per worker in large manufacturing firms under risky or growing conditions. As the productivity of older workers may vary, future research may determine under what conditions – size, industry, region and financial conditions – older workers contribute positively to productivity. Firms with financial troubles or those planning to downsize should be cautious about laying off older workers as an approach to improving organisational performance because these workers contribute positively to productivity under certain conditions.

Introduction

Population ageing is a global phenomenon associated with many socioeconomic changes. The combination of lower fertility, better health care and longer life expectancy has contributed to population ageing in many developed countries. As one example, the proportion of employees aged 55 or more is increasing in the United States (11.9% in 1990, 13.1% in 2000, 16.2% in 2005, 19.5% in 2010, 22.1% in 2015 and an expected 24.3% in 2020) (Toossi and Torpey, 2017). Most governments are feeling the effects of budget deficits caused both by low tax revenue from a shrinking workforce and high expenditures on pensions and health care for the aged. Reduced consumption, the propensity to save for later life and investment of savings in safe assets have also contributed to the challenge faced by governments. When low consumption and investment are coupled with a shortage of workers, difficulties in economic growth may follow.

In the future, population ageing will continue to be a significant challenge. Figure 1 shows median ages in the world from 1980 to 2050. Table 1 ranks the world’s oldest countries in 2015, 2030 and 2050. The average median age of the world’s population was 22.5 years in 1980; this will increase to 36.1 years by 2050. The median age of the population in European Union (EU) of 15 countries¹ will also increase from 33.1 years to 46.9 years. The most dramatic change in population structure, however, will happen in East Asian countries. Looking at Korea as an example, the median age of the population was similar to the global average in 1980. However, since 2003, South Korea has had the lowest fertility rate among all Organisation for Economic Co-operation and Development (OECD) countries. In 2050, South Korea will be the oldest country in the world. More than one-third of the Korean population will be over the age of 65 and around half of all workers will be aged 50 or over.

Figure 1. Median ages of South Korea, Japan, EU15, USA and World. Data source: World Population Prospects: The 2015 Revision, Department of Economic and Social Affairs of the United Nations, 2015.

Table 1. Rankings of oldest countries by median age.

Data source: World Population Prospects: The 2015 Revision, Department of Economic and Social Affairs of the United Nations, 2015.

‘Not specified’ means countries or areas with fewer than 90,000 persons.

To cope with the challenges of population ageing, the OECD recommended that the Korean government provide work incentives and opportunities for older people (Keese, 2004) and that regulations regarding the retirement age of the workforce should be modified. In addition, the OECD encouraged private companies to enhance job opportunities for older workers. Along with these recommendations, older Korean workers are now encouraged to participate in up-to-date training and to have good access to employment services and working conditions considering their physical limits.

Population ageing brings into focus the need for greater employment of older workers; however, this raises questions about their productivity. What can be done is to find out whether they are beneficial to a company’s productivity or are as productive as young workers. In that sense, the productivity of older workers should be examined more thoroughly.

As previously mentioned, South Korea is the fastest-ageing country in the world. From 1980 to 2050, South Korea will have changed from one of the world’s youngest countries to the oldest. In addition, South Korea boasts a variety of companies, from large conglomerates (i.e. Chaebols) to small startups (e.g. information technology (IT)-based game companies), from the entertainment industry (e.g. the Korean wave) to the heavy chemical industry (e.g. ship-building). Korean companies provide a well-diversified sample for firm-level research on the issue of ageing.

It is widely believed that ageing may have a negative effect on productivity. In some industries or workplaces, however, older workers may provide more benefits to firms than younger workers due to their valuable experience. In this article, therefore, we analysed the relationship between older workers and productivity using Korean workplace survey data to explore the benefits of employing older workers from multiple perspectives.

What have previous empirical studies found about age and productivity?

There are many papers on the relationship between ageing workforces and outcomes for organisations. Several characteristics caused by ageing can be positive for an organisation,⁶ while others can be negative; in turn, organisations may have outdated ideas about older employees or difficulty establishing effective policies (Posthuma and Campion, 2009). Typical stereotypes about older employees are as follows: that they have low motivation, little desire for training, resistance to change, low trust, bad health and excessive dedication to family (Ng and Feldman, 2012). Some studies suggest that because older workers may be less concerned with career progression, their level of effort may seem inadequate (Rabl, 2010; Wong et al., 2008). Other studies suggest that older workers may be less inclined to undertake training (Greller, 2006; Maurer et al., 2008). Also, the belief that they have less capacity to learn (Wrenn and Maurer, 2004) and are difficult to teach (Fritzsche et al., 2009) are aspects of this stereotype. Older employees are also seen as resistant to change or as having less capacity to change (Morris and Venkatesh, 2000). Other studies have suggested that older employees are less open than younger workers to adopt new technology (Davis and Songer, 2009; Meyer, 2011) and they are less adaptable to the learning of new tasks and elements of different cultures (DeArmond et al., 2006). The fourth stereotype is that older workers are less trusting and as a result, more isolated (Pinquart and Sorensen, 2001; Victor et al., 2002). Also, lower interpersonal skills (Bal et al., 2011) and lower-quality service to customers (Luoh and Tsaur, 2011) have been reported. Older workers may also experience more health problems at work (Kite et al., 1991). Young employees’ view of old employees is that they stereotypically have mental health problems and that they are miserable, ill-tempered and complaining (Hummert, 1990). Finally, they are assumed to be more dedicated to their family, or more interested in leisure activities and travel, rather than being focused on their work (Gordon et al., 2007; Ng and Feldman, 2007).

In a meta-analysis, Ng and Feldman (2012) examine these stereotypes in 418 empirical studies, investigating whether older employees really have such negative characteristics or not. Their conclusion is that all the stereotypes mentioned above are wrong, except the finding that older employees do have less willingness to participate in job training or career development activities. Older workers are not less motivated, more resistant to change, less trustful or in poorer health and the imbalance between work and private life is minimal. Meanwhile, several studies insist that older workers have strengths. Ng and Feldman (2008) show that age is unrelated to core task performance and creativity. Furthermore, in some respects, older employees perform better than young employees in organisations; they are more involved in organisational citizenship behaviour (OCB), which means helping others without expecting rewards (Organ, 1988), and display fewer counterproductive work behaviours, that is, behaviours that negatively affect well-being (Rotundo and Sackett, 2002). Also in the last-named study, age is negatively related to tardiness and absenteeism. Ng and Feldman (2010) identify the following advantages of retaining older workers: high job satisfaction, interpersonal satisfaction and greater commitment. In short, the characteristics of older workers differ from those of younger workers; some are positive and some are negative. Thus, we can infer that hiring older workers may have different effects on productivity within firms depending on the characteristics or circumstances of the firm.

From the perspective of productivity, the results of the aforementioned studies are not consistent. It seems that there is a stereotype that the productivity of older employees is lower than that of younger employees (Van Dalen et al., 2010). Frosch (2011) shows in his literature review an inverted U-shaped relationship between age and innovative performance in terms of patent applications (Henseke and Tivig, 2008), great inventions (Jones, 2010; Lehman, 1966) and research productivity (Pelz and Andrews, 1966; Stahi, 1977; Vincent and Mirakhor, 1972). Similarly, in an analysis including data from the International Labour Organisation (ILO) and Penn World Table version 6.0, Feyrer (2007) finds that hiring workers aged 40–49 years has a large positive effect on aggregate productivity. Lindh and Malmberg (2009) test the correlation between age structure and economic growth in 15 EU countries, and find that an increase in workers in the 50–64 age group has a positive effect on gross domestic product (GDP) per capita, but an increase in workers over 65 years has a negative effect. Vandenberghe et al. (2013) found a negative effect of hiring older workers on the productivity – labour cost gap; their lower productivity is not compensated by low labour costs. Hellerstein et al., (1999) found that for workers aged 35–54 and those aged over 55, productivity and earnings rise at the same rate, but their research was unclear about differences in productivity by age.

Studies specific to countries, industries and firms of different sizes have provided mixed results. Malmberg et al. (2008) found a positive correlation between age and productivity in Swedish Manufacturing and Mining Surveys from Statistics Sweden. Lallemand and Rycx (2009) in a Belgian study found that firm productivity decreases substantially as workers get older and that the negative effects are much stronger in information and communication technology (ICT) firms than in non-ICT firms. Van Ours and Stoeldraijer (2011) argue that there is no evidence of an age effect on the pay – productivity gap in Dutch manufacturing firms. Mahlberg et al. (2013) concluded that the relationship between age and productivity differs across industries, regions and firm characteristics and that differences in productivity can be caused by different characteristics of older employees.

Han and Suen (2011) assert that young employees have a greater tendency to switch to different industries. Analysing the age composition of 25 industries in Hong Kong, they found that employees behave differently depending on whether the industry is growing or declining and that organisations hire older employees or younger employees for different reasons. Agriculture or fishing-related industries prefer older workers, while communication and finance-related industries prefer younger workers. They also find that education level and gender have effects on average age in specific industries and that an increase in certain industries’ share of the labour market results in a decrease in the average age of employees in that industry.

Given these varied findings, we seek to establish which characteristics of older workers are associated with greater productivity, and under which circumstances. In this study, we analyse the effects of age on productivity with consideration of several elements such as industry, financial condition and firm size and investigating the conditions under which hiring and retaining older employees are associated with high productivity.

Data and theoretical framework

In this article, we gathered data from the Workplace Panel Survey (Korea Labour Institute, 2017), approved by the Korean government and governed by the Korea Labour Institute, a government-funded policy research body. The Korea Labour Institute has collected data through this survey every 2 years since 2002. The sample includes workplaces across the country with 30 or more employees. The survey includes demographic information (number, age, sex and position of employees), financial performance information (balance sheet, income statement and so on), workplace characteristics (major stockholders, foreign/domestic owner, innovations and so on), employment management (recruiting, education, promotion and so on), compensation, union–management relations and so on. Workforce age information was included only in the 2007, 2009, 2011 and 2013 surveys; therefore, we used the data for these years.

Dependent variable

In our empirical analysis, we measure a log of value added per employee as the dependent variable. This variable was previously utilised in the study of Malmberg et al. (2008) to measure labour productivity. We define value added as the sum of earnings before tax, labour costs, net financial costs, rent, tax and duties and depreciation expenses. The Workplace Panel Survey provided value added data for each company.

Independent variable

To determine the effects of age on productivity, the following age groups were specified: under 30 years, 30–49 years and over 50 years. This division has been used in several other studies analysing the effect of age on the economy (Lindh and Malmberg, 1998, 1999, 2007; Malmberg et al., 2008; Mahlberg et al., 2013).

Control variables

We utilise the Cobb-Douglas production function. This function is the most commonly used productivity model in economics papers. In it, productivity (Y) is expressed as a function of labour (L), capital (K) and other factors (A), as follows

$\begin{array}{c}Y=L^p{K}^{q}A\\ \frac{Y}{L}=L^{\left(p+q-1\right)}{\left(\frac{K}{L}\right)}^q\end{array}$

Therefore, we should include the number of workers and capital per worker among the control variables.

According to the protocol in previous studies (Malmberg et al., 2008), education level was measured as the number of years of education and used as a control variable. Work experience was also used as a control variable consistent with Mahlberg et al. (2013), who used tenure as a control variable. The second control variable is firm size; this was also used as a control variable by Malmberg et al. (2008); it was represented as labour variable in our article. The third variable is small or medium-sized enterprise (SME); whether a firm is SME or not. Firms with fewer than 300 employees were classified as SMEs according to the Korean government standard. It is necessary to control for this variable because it is related to the benefits of SMEs, such as tax incentives and mandatory government procurement.

Estimated model

The following is the model that was estimated in this study

$Y_{it}=\alpha +{\beta }_0\times {\left(\mathrm{Ratio}\mathrm{of}\mathrm{workers}\mathrm{aged}\mathrm{over}50\mathrm{to}\mathrm{total}\mathrm{workers}\right)}_{it}+\sum _{k=1}^n{\beta }_k{X}_{it}+{\epsilon }_{\mathrm{it}}$

where the outcome variable $Y_{it}$ is the log value of value added per worker of the i-th firm nested in year t and $X_{it}$ contains the control variables of the i-th firm in year t. These control variables are as follows: the ratio of workers aged under 30 to total workers, the log value of number of workers, the log value of tangible and intangible assets per worker, the average education level of workers, the average total number of months the workers have worked for the organisation, the square of this average, an industry dummy, a year dummy and an SME dummy.

Tangible assets are assets with physical forms and durability measured over 1 year. They include land, buildings, machinery, vehicles and other things owned by the organisation. Intangible assets are non-physical assets such as patents, trademarks and copyrights. The average education level is a discrete variable, calculated as follows: 1 means that the average education level of all workers is less than middle school, 2 means high school, 3 means they have completed 2 years of college, 4 means they have completed 4 years of college and 5 means that their average education level is higher than a master’s degree. The number of employees were used as labour variable. SME is a dummy variable, which takes a value of 1 if the number of workers is less than 300 and 0 otherwise. The number 300 comes from the Korean legal standard by which companies are categorised into large companies or SMEs.

Results

Results of the main analysis

Table 2 shows the sample companies included in the analysis. ‘Other industries’ includes all industries except the manufacturing industry; it includes service and IT industries such as retail, information and technology, finance and transportation. SMEs and large companies were categorised by the number of employees. According to the Korean Framework Act on SMEs of 2013, companies with fewer than 300 employees were classified as SMEs. Therefore, in this study, companies with more than 300 employees were considered to be large companies. The total number of companies in the dataset was 7017. Of these, 2090 firms were considered as manufacturing SMEs and 831 firms were large manufacturing firms.

Table 2. Number of companies in data sample, Republic of Korea.

SME: small and medium sized enterprise.

SME – less than 300 workers, large companies – more than 300.

Table 3 displays the ratio of workers aged over 50 to total workers in the sample companies. The ratio of older workers increased rapidly as time went on. In 2007, the mean ratio of all companies was 0.142. However, in 2013, the mean ratio increased to 0.224. SMEs stand out in this workforce ageing process. In both manufacturing SMEs and SMEs in other industries, the mean ratio of workers aged over 50 increased by about 10% from 2007 to 2013.

Table 3. Mean ratio of workers aged over 50 years in 2007, 2009, 2011 and 2013.

SME: small and medium sized enterprise.

Table 4 presents correlations between variables. The ratio of workers aged over 50 had a negative correlation with value added per worker, while the ratio of those in their 30s and 40s had a positive correlation at the 99% confidence level. However, no significant correlation was evident for the ratio of workers under 30. These correlations between age distribution of workers and their productivity reveal very primitive relations among variables, which coincide with the results of previous research. In the following sections, we attempted to verify these relations by firm size and financial conditions of firms.

Table 4. Pearson correlations between variables.

SME: small and medium sized enterprise.

p values in parentheses,

+ p < 0.1,

* p < 0.05,

** p < 0.01.

Table 5 presents the results of the regression analysis for the proportion of employees over 50 to logged value added per worker. For all models with or without control variables and with or without fixed effects, the effect of this group on value added per worker was negative and statistically significant. In Model 8, in which all related variables were included, an increase in the proportion of workers over 50 by 10% reduced value added per worker by 3.38%. Therefore, we may initially conclude that workforce ageing has a negative effect on labour productivity. This result coincides with those of previous research that explore the relation between workforce ageing and productivity. However, the proportion of workers under 30 also had a negative relation with value added per worker. While the coefficient of working experience was positive, the coefficient of working experience squared was negative.

Table 5. Regression results of baseline model.

DV: dependent variable; FE: fixed effect.

p values in parentheses,

+ p < 0.1,

* p < 0.05,

** p < 0.01.

Though low productivity of firms due to ageing of workers has been addressed in previous studies, we focused on the possibility of positive or non-negative effects of retaining older workers on productivity in this article. In previous research, the most commonly used classifications were industry, size and financial condition (Becker and Dietz, 2004; Han and Suen, 2011; Mahlberg et al., 2013). Therefore, we analysed older workers’ productivity using these classifications.

First, we hypothesised that older workers, in the manufacturing industry, contributed more to firm productivity than younger workers, because the former were more likely to be skilled than the latter and experience plays a more important role than newer knowledge. As managers or intermediate managers, older workers can transfer their know-how to young and unskilled employees. In other industries such as IT and finance, on the other hand, we supposed that new knowledge is as important as or more important than experience. In Models (1) and (4) in Table 6, however, the effects of older workers on productivity were negative. The effects were non-negative only in Model (3).

Table 6. Regression results analysed by industry sector and firm size.

DV: dependent variable; FE: fixed effect; SME: small and medium sized enterprise.

p values in parentheses,

+ p < 0.1,

* p < 0.05,

** p < 0.01.

Second, we hypothesised that older workers’ characteristics would have different effects on productivity levels according to companies’ financial conditions. This idea was inspired by the findings of Han and Suen (2011). When industrial restructuring was implemented in Hong Kong, declining sectors such as agricultural products tended to retain older workers. Likewise, it may be that for companies that undergo big changes or are in financially dangerous situations such as nearing bankruptcy, older workers’ OCB and less counterproductive work behaviours may be beneficial to company productivity. Older workers may, therefore, be helpful and enhance organisational stability during challenging times.

As seen in Table 6, we divided our sample into two subsamples by firm size. Interestingly, large manufacturing firms were not much affected by workforce ageing, while the coefficients for SMEs were negative, corresponding to the proportion of workers older than 50. Therefore, as hypothesised, in manufacturing industries, we found a positive relationship between ageing and productivity in larger firms. To explore these non-negative effects of ageing more thoroughly, we used firm’s financial situation as well as firm size in supplementary analyses described in the following subsections.

Risky companies

We divided the sample again into two groups according to their financial situation, hypothesising that the productivity of older employees would differ between firms in different financial states. To verify how the effect of ageing on productivity differs by financial condition, we compared two groups of firms: financially safe and risky firms.

We defined risky companies as those unable to repay their interest expenses using their annual income. If a company’s operating income is less than its interest expenses, the company is considered as risky and financially unsustainable. Risky companies are usually driven to bankruptcy due to cash flow problems. According to the standard set by the Bank of Korea, if the interest coverage ratio (this equals the operating profit over the interest expense) of a company is less than 1, the company is considered as risky.

We then analysed the relationship between ageing and productivity in the manufacturing sector. Table 7 indicates the results of the regression analysis of both risky companies and safe companies. Models 1 and 2 included the subsample of SMEs and Models 3 and 4 included the subsample of large companies. Like the baseline model, in Model 2, an increase in the proportion of older employees by 10% had a negative effect on value added per worker, reducing the value for this variable by 3.33%. However, in Model 3, an increase in the proportion of older workers by 10% had a positive effect on the dependent variable, increasing value added per worker by 13.08%.

Table 7. Regression results analysed by firms’ financial situation (‘risky’ or ‘safe’).

DV: dependent variable; FE: fixed effect; SME: small and medium sized enterprise.

p values in parentheses,

+ p < 0.1,

* p < 0.05,

** p < 0.01.

A firm is defined as ‘risky’ if its annual operating income is less than its interest expenses.

When we compared the effects of retaining older workers on value added per worker by firm size, values for larger firms were more positive for risky firms and less negative for safe firms. Hence, we can conclude that retaining older workers in large firms is not actually detrimental to firm productivity. Moreover, the results in Table 7 imply that retaining older workers in risky firms is also less detrimental; we reached this conclusion by comparing the results in Models 1 and 3 with the results in Models 2 and 4. Risky companies seem to be in a dangerous situation, nearing bankruptcy and characteristics of older workers such as engaging in OCB may help to stabilise the nervous atmosphere within an organisation, which can enhance productivity. Börsch-Supan and Weiss (2016) provide similar explanations. According to their research in a truck assembly plant, older workers are less prone to make serious errors. This suggests that older workers have good capabilities for figuring out difficult situations and improving overall productivity by focusing on vital tasks.

Firm growth

Next, we split the data into two subsamples based on the growth rate of firms. We defined growing companies as those with a sales growth rate that is larger than average. We divided manufacturing companies into four groups by firm size and growth rate, as shown in Table 8. The subsamples of Models 1 and 3 (or Models 2 and 4) are firms of growing (or not growing) status, respectively, for which the growth rate was greater (or less) than the average growth rate. More specifically, the growth rate of firms for Model 1 (or 2) was greater (or less) than the average growth rate among SMEs in the entire sample. Firms for Model 3 (or 4) were chosen when the growth rate was greater (or less) than the average among large firms.

Table 8. Regression results analysed by firms’growth status.

DV: dependent variable; FE: fixed effect; SME: small and medium sized enterprise.

p values in parentheses,

+ p < 0.1,

* p < 0.05,

** p < 0.01.

Table 8 provides the results of the regression analysis, which imply that retaining older workers can be beneficial to firm productivity. Retaining workers aged over 50 in growing SMEs had a negative effect on value added per worker, but in large, growing companies it had a positive effect. These results also confirm that large manufacturing firms can retain ageing workers without decreasing productivity.

Profitable companies

Finally, we considered profitability. A profitable company is one whose profit ratio is larger than average, where profit ratio is defined as the ratio of operating profits to sales. We divided manufacturing companies into four groups: profitable SMEs, unprofitable SMEs, profitable large firms and unprofitable large firms. Profitability was determined by comparing the profit ratio of a given firm with the average profit ratio among firms of the same size.

Table 9 lists the results of the supplementary analysis, which are consistent with those in Table 8 indicating that hiring or retaining older workers has a less harmful effect on productivity in large firms. As shown in Table 9, there was no significant effect of an increase in older workers on value added per worker in large firms, while the effect of retaining older workers on SMEs was negative. However, the findings are consistent with the results above in that they demonstrate that an ageing workforce can be less harmful in large manufacturing companies.

Table 9. Regression results analysed by manufacturing firm size and profitability.

DV: dependent variable; FE: fixed effect; SME: small and medium sized enterprise.

p values in parentheses,

+ p < 0.1,

* p < 0.05,

** p < 0.01.

Robustness check

In order to check the robustness of our model, we conducted a residual analysis, creating residuals by using our model. Figure 2 is a residual plot. The X-axis represents the ratio of workers aged over 50 to total workers in each company. The Y-axis represents residuals of our top model. We found that the sum of the residuals is almost zero ( $-5.78\times {10}^{-10}$ ).

Figure 2. Residual plot.

We also checked for endogeneity problems on the right-hand-side variables. Our literature review showed that for most variables on the right-hand side – control variables such as capital per worker, education level, work experience, industry, year and labour – no serious endogeneity problem was evident. On the other hand, the independent variable – the ratio of older workers – is an important variable; therefore, we addressed potential endogeneity problems with the independent variable using instrumental variables. Table 10 presents the results of the regression analysis testing for endogeneity using the two-stage least squares (SLS) method instrumented by the lagged proportion of workers over 50 years. The results of the 2SLS analysis are similar to those of the ordinary least squares (OLS) analysis. Based on these results, we can say that endogeneity is not a serious problem in this analysis. We also found similar results when considering the re-employment system, that is, discharging and rehiring older workers as temporary workers. This factor directly affects the proportion of older workers because firms in which a re-employment system operates may have a low proportion of older workers; therefore, it is an appropriate instrumental variable.

Table 10. Regression results of endogeneity test.

DV: dependent variable; FE: fixed effect; SME: small and medium sized enterprise.

p values in parentheses,

+ p < 0.1,

* p < 0.05,

** p < 0.01.

Table 11 provides the results of the regression analysis with instrumental variables included (2SLS). The coefficient of the proportion of workers over 50 years instrumented by the re-employment dummy was much larger than the coefficient instrumented by the lagged variable. When the re-employment instrumental variable was used, an increase in the proportion of older workers by 10% reduced value added per worker by 12.65%. When the lagged variable was used, value added per worker was reduced by 5.89%.

Table 11. Regression results with instrumental variables.

Note: p values in parentheses,

+ p < 0.1,

* p < 0.05,

** p < 0.01.

Discussion and conclusion

In this article, we have explored the effects of hiring or retaining older workers on productivity in companies. On the whole, an increase in older workers had a negative effect on value added per worker in the firms in our sample. This was consistent with most findings in previous research conducted using European or Taiwanese data.

However, the results were the opposite when companies’ financial condition and firm size were taken into account. In large manufacturing companies, an increase in older workers had positive effects on value added per worker for risky or growing companies. As for manufacturing SMEs, an increase in older workers had negative effects on value added per worker under most financial conditions. Figure 3 presents the range of coefficients for the proportion of older workers. In large manufacturing companies under risky or growing conditions, the coefficients were within the positive range; though in negative ranges in other cases. Therefore, employers should be cautious about laying off older workers.

Figure 3. Coefficients of the proportion of workers aged over 50.

Some employers may think laying off older workers is good for their company’s performance, based on inaccurate stereotypes. However, they should consider the conditions of their organisation first and foremost as, under certain conditions, retaining older workers can be beneficial. Older workers are beneficial (or at least less likely to be harmful) to firm productivity under specific conditions. They seem to fit well in large Korean manufacturing firms, for example, in terms of organisational structure, values and culture. Among the various characteristics of large Korean manufacturing firms, some characteristics are derived from Korean traditional values and culture. Shin (1992) has suggested the following characteristics of Korean corporate culture: anthropocentric, paternalistic, elder-centric, collectivistic and hierarchical. Cho et al. (2014) identify dynamic collectivism as the common characteristic of Korean large companies. Such corporate cultures seem to fit well with older workers. Older workers often demonstrate strong commitment to companies and generate interpersonal well-being (Ng and Feldman, 2010). They easily accept organisational culture and are willing to follow the rules. A good fit between older workers and their organisations contributes positively to firm productivity.

Our research is distinguished from previous research in several respects. First, we explored the ageing–productivity relationship by analysing balance sheets and income statements to determine the financial condition of the firms in our sample. Han and Suen (2011) studied age structure in growing and declining industries. They define ‘growing’ and ‘declining’ based on changes in employment share. We clarify the definition of ‘growing’ or ‘declining’ by analysing each company’s balance sheet and income statement. We divide companies into risky and safe companies based on their interest coverage ratios (operating income/interest expenses). Also, we divide companies into growing/not growing companies and profitable/unprofitable companies by analysing sales growth and operating profits. Second, we utilise not European data, but data from East Asia, where the population is quickly ageing. Though the population of East Asian countries is ageing faster than in any other region, very few research papers have used East Asian data. As presented in Table 1, in 2050, four East Asian countries – South Korea, Japan, Singapore and Hong Kong – will be among the top 5 oldest countries in the world. As the number of studies based on fast-ageing East Asian countries is minimal, analysing Korean data is valuable.

There are several directions for future research. In addition to academic researchers, policy makers need to study the circumstances under which older workers are more productive. As the productivity of older workers differs by industry, region and firm characteristics (Mahlberg et al., 2013), future studies should include diverse countries, industries and other characteristics. The results of such research could be directly helpful to governments, who should consider them when making policies. In future research, the challenge of ageing workforces and its relation to the fourth industrial revolution may be an interesting focus. In addition, automation and robots may help older workers increase productivity, or they may replace the workforce altogether. Meanwhile, the fourth industrial revolution may completely change the definition of productivity. Future researchers can study this and the productivity of older workers together. The results of such research may be meaningful to many stakeholders.