IMPACT OF AGE ON PRODUCTIVITY The changing scientific and technological policies and differing socio-economic environments have generated

renewed interest in the productivity of scientific community worldwide. One such concern is that whether the aging of scientific community will affect the nations rate of scientific advances and scientific productivity of researchers. The scientific productivity of an individual scientist is generally measured in terms of various scientific and technical outputs. The research output of individual scientist is also affected by various factors such as social, psychological and economic, which are interrelated in a complex manner. Relation with age and productivity also depends on the research involvement of institutional setting, reward system, and generally socioeconomic environment of a given country. A high share of older employees can be beneficial for the productivity of establishments because on average older employees have more experience, are more loyal, and have a higher quality conscience and working morale. On the other hand, older age is frequently associated with less technical knowledge, creativity, innovativeness, flexibility, openness to new knowledge, and physical as well as psychological resilience psychologists and medical scientists argue that so-called fluid cognitive abilities such as the performance and speed of solving tasks related to new material tend to deteriorate with age. Other,crystallised abilities such as felicity and word fluency improve with accumulated knowledge and remain at a high functional level at least inside the age-bracket that is relevant for employment occupations that are likely to become less productive when the employees age. Ageneutral occupations are those of bank or commercial clerks and electronic engineers. Finally, examples for occupations that might even have a higher productivity when employees are old are lawyer, professor, manager, medical doctor or engineer. However, no clear hypothesis on differences in relative productivity between age groups in different sectors can be drawn from these findings because in most sectors occupations with increasing and decreasing productivity potential are employed. Production in the manufacturing sector is frequently characterised by a relatively rigid and monotonous pace dictated by the conveyor belt (Berg, 1994) and is often physically demanding. The same applies to metal manufacturing where conveyor belts in the production sector are

very common. However, jobs in the service sector are often characterised by the absence of physically demanding production methods. They are frequently psychologically demanding, however, because the individual output is easy to measure, such as in the banking and insurance industry. Jobs in the service sector often imply a high degree of social interaction and communication skills. A priori, we expect a decline in the relative productivity of older workers in the production sectors, while we expect the productivity for the service sector to decline at a slower pace. The technology and production processes should determine to which degree human capital investments drive the productivity contribution of employees. If production requires a lot of firm or sectorspecific human capital, one would expect to have a steep learning curve, particularly for young employees who enter the labour market and on whom training investments are concentrated. Since the manufacturing sector typically requires more specific knowledge than the service sector (Mohrenweiser and Zwick,2009), we expect a steeper learning curve in this sector. More specifically, the relative productivity for young workers in comparison to old and prime age workers is likely to be lower in the manufacturing sector. The higher importance of continuing training in addition might reduce the relative productivity of older employees in the manufacturing sector older employees take part in training less often. An important reason for the decline in training participation over the life cycle might be that personnel managers perceive older employees to be less able and willing to learn (Boockmann and Zwick,2004). When older employees take part in training, their participation seems to be less effective than for younger employees (Gbel and Zwick, 2010). Therefore, we would expect a decline in relative productivity of older employees in sectors that need continuous training efforts.Many contributions stress the impact of information and communication technology (ICT) on labour productivity. For example, firms in the service sector frequently use ICT more intensively than manufacturing enterprises (O Mahony and Van Ark, 2003). Older employees being less able to cope with the specific demands of ICT usage might reduce the relative productivity of older employees in ICT intensive enterprises (Lallemand and Rycx, 2009; Bertschek and Meyer, 2009). Despite the introduction of ICT, there are many service sectors where technological shocks play

only a minor role, however, since the core of the production processes remains virtually unchanged over time. On the contrary, ICT has been introduced in selected production processes in manufacturing. From a theoretical point of view, the role of technology shocks on the ageproductivity profiles of service or manufacturing sectors is therefore not determined

workforce age to performance treat performance as one-dimensional and often focus on individual not organizational performance. To analyze the effects of workforce age on organizational performance, we suggest treating performance as multidimensional with at least two output dimensions: quantity and quality. We provide a novel conceptual framework that borrows insights from multiple disciplines to better understand ageing phenomena in organizations. In particular, we build on psychological and medical studies showing that individual age has different effects on different cognitive capabilities. As a result, we argue that workforce ageing may affect various performance dimensions, such as quantity and quality, in different, often opposite ways. In our empirical part we examine a unique dataset containing detailed court data over a time span of 19 years. We find that average workforce age is linked negatively to quantitative organizational performance but positively to qualitative organizational performance. Our findings suggest that future organizational studies should decompose performance at least along the quantityquality dimension. Our theoretical framework helps to understand the different types of ageing effects and to derive itemized implications for changes in organizational performance. It also helps to reconcile contradictory findings of previous studies and to derive important managerial implications for task assignments, career policies and company strategies in view of upcoming demographic changes in many developed countries.Entrepreneurial activity, start-up of new firms or expansions of existing ones, are more likely to be carried out by relatively young adults, according to findings from the 34 countries surveyed in the Global Entrepreneurship Monitor (GEMConsortium, 2004). Peak

entrepreneurial activity is found among individuals aged 2544. In support of a young entrepreneurial peak, Aubert et al. (2007) find, in a French study, that the more innovative a firm is, the higher the wages of younger employees are relative to wages of other age groups. In order to describe age variation in performance of the average worker rather than elite groups, one needs a different set of productivity measures. Supervisors ratings are often used to identify the relation between the employees age and his or her productivity. McEvoy and Cascio (1989) review 96 studies on the impact of the employees age on supervisors assessment and sales records and find no clear effect of age on productivity. Likewise, reviews by Warr (1994) and Waldman and Avolio (1986), based primarily on supervisor assessment, find no or a slightly negative impact of age on job performance. However, Remery et al. (2003) find that older individuals are seen as less productive in particular in firms with a higher proportion of senior workers, which is where knowledge about older individualswork capacities is likely to be highest, in a survey 1007 Dutch personnel managers. A general problem with most approaches used to measure age variation among workers is the sample selection problem. Good workers get promoted, while inefficient workers may lose their jobs. Hence, positive selection can increase by age, which could lead to bias in the estimates of the oldest age groups working capacity. Self-assessed work ability is sometimes considered, and estimates from Finland suggest a clear decline by age of workers (van Ours et al. 2007). Solem (2006) presents evidence from Norwegian surveys on subjective general work ability that shows a decline from the 30s to the 60s. Nevertheless, respondents believe that they are equally capable of performing their work over the same age interval; although general work ability declines with age. This may be because relevant abilities are fully maintained and the worker is increasingly well matched to the type of work task he/she performs. A particular problem with managers ratings and selfevaluations is that these evaluations can be biased as they are subjective. Evaluations of older workers could therefore be inflated due to loyalty concerns and as a reward for past achievement or in the case of self-reports, as a means of self-justification, or even selfdegradation. Discriminatory attitudes among managers towards older or younger individuals can also affect an age-productivity estimate, which reduces the validity of this approach (Levy,

2003; Salthouse and Maurer, 1996).Measuring the impact of age on job performance is sometimes based on measures of the quantity and quality of a workers output. Studies based on this approach tend to find that older employees have lower productivity levels. A study of several industries from the U.S. Department of Labor (1957) finds that job performance increases unti the age of 35 and steadily declines thereafter. At the end of the career, productivity declines by 14% in the mens footwear industry, and 17% in the household furniture industry. Czaja and Sharit (1993) investigate the extent to which age has an impact on computer-based work performance and find that increased age was associated with longer response times and a greater number of errors for all tasks considered. These task-quality/speed tests are potentially more objective as they do not rely on subjective assessment. However, they may be biased from the fact that the workers are selected in terms of age groups and occupational types (Rubin and Perloff, 1993). Further, the time-limit common in such studies may bias results. For example older employees may maintain a higher work speed in the short period they are studied than what they would be able to do in a normal job situation (Salthouse and Maurer, 1996). A recent estimation approach is based on analyses of employer-employee datasets (e.g., Gelderblom and de Koning, 2002; Hgeland and Klette, 1999). A key strength of these estimates are the large samples, which can encompass most workers in some of the main industries in a country; several samples include a few million individuals and thousands of firms. The most common finding from these studies is a hump-shaped relation between job performance and age. An overview over analyses of the effect of age on productivity using employer-employee datasets are given in the Annex. Of the 14 studies considered, 11 find a productivity decline in the 50s relative to the 30s and 40s, two have inconsistent results, while one finds that productivity peaks among the oldest workers. However, bias may come from the fact that many determinants of a firms value-added, such as capital levels, are either omitted or poorly measured. The reverse causality problem could also be an issue a companys success can increase the number of new employees and lead to a younger age structure, which could mean that a young age structure could be the consequence rather than the cause of a companys success. Wages are often determined by other factors than

individual contribution to the firms value-added, including the role of unions, uncertainty about new workers productivity and delayed payment contracts where performance while being young results in higher earnings for those still employed while they are older (Agell and Lundborg, 1995; Freeman, 1982; Hutchens, 1989). Ageearnings profiles can, however, provide information on productivity differences in settings where wages are likely to reflect actual productivity. One example is a study by Lazear and Moore (1984) who examine the difference between earnings profiles of the self-employed and salary workers. Kotlikoff and Gokhale (1992) study agevariation in the wages offered when firms hire new employees. Their findings suggest managers have earliest productivity maximum, while office and sales-workers have somewhat older peaks, though they all reach their highest productivity in the 40s, with sharp declines thereafter. Age diversity has been suggested to have a positive effect on productivity through better age-matching between salespersons and customers and through complementarity of different age groups, but the evidence seems inconclusive. Experimental research suggests cooperation is highest when juniors and seniors are mixed together A final point that has hardly been investigated until now is that human resource management measures vary among different sectors. For example, the application of specific measures that are targeted on older workers could vary significantly between different sectors. Weichel et al. (2008), for instance, report that measures specifically aiming at improving the relative productivity of older employees are more frequently used in (metal) manufacturing. To summarise our considerations so far, there are many

theoretical arguments that suggest that the age-productivity profiles of manufacturing and service sector might differ. The extent and the direction of the differences remain an empirical question, though. As metal manufacturing is one of the most important economic sectors in Germany and frequently has been in the focus of the discussion concerning negative impacts of an ageing workforce (Weichel et al., 2008), we estimate the age-productivity profiles for this sub-sector as well. During recent years, there has been a growing interest in estimating the causal relationship between workforce age and establishment productivity. Brsch-Supan et al. (2005), Skirbekk (2008), and Gbel and Zwick (2009) provide extensive literature surveys of this quickly emerging field. In the following lines, we therefore only briefly review the relatively scarce empirical literature on differences in the impact of ageing workforces on establishment productivity between economic sectors. Crpon et al. (2003) split their sample into the French

manufacturing and non manufacturing sector and drop the construction sector. They hardly find any differences in the relative impact of age groups on productivity for both sector groups besides a larger productivity disadvantage of young employees in the manufacturing sector than in the non-manufacturing sector. The authors therefore do not discriminate between the sectors in the discussion of their results. Summing up, the empirical literature so far suggests that there might be differences in the age-productivity profiles between sectors younger employees seem to be less productive in the manufacturing than in the services sector because specific human capital acquired in the job seems to be more important here. ICT usage seems to negatively affect the relative productivity of older employees mainly in the service sector. The results are frequently not very robust with respect to changes in the estimation technique and tend to disappear in estimates that control for unobserved heterogeneity or endogeneity of the age shares in production functions.