Productivity and Employability of older workers

Transcription

1 Productivity and Employability of older workers Evidence from Belgian Firm-Level Data V. Vandenberghe, M. Rigo, F. Waltenberg (IRES-ESL-UCL) Immaq Workshop on Pension and Ageing Jan. 27, 2015 UCL-Louvain-la-Neuve

2 Presentation outline 1. Context/Motivation 2. Our focus : old labour demand 3. The existing literature 4. Methodology/identification strategy 5. Data 6. Conclusions & policy implications 2

3 1. Context, motivation - Ageing population, political initiatives to increase older empl. rates that are very in some EU countries (e.g. Belgium, France, Luxembourg) - Existing literature looks mainly at the consequences of an ageing population, in terms of welfare cost or growth (Gruber and Wise, 2004) the retirement behaviour of older individuals (replacement rates, pension, early-retirement schemes, role of health, jointdecision within households ) (Mitchell & Fields, 1983) <=> supply side - Not so much the determinants of the labour demand by firms (e.g. labour costs, productivity...) <=> demand side 3

4 Although evidence that it may matter abounds A high share of involuntary retirement (55-64) Source: ISSP

5 Belgium 5

6 2. Our focus Are employers willing to (re)employ older workers? analyse firm-level panel data to assess the impact of larger shares of older workers on [a] Labour Productivity [b] Labour cost (gross wage+ social sec. costs) [c] Productivity/labour cost ratio => Key assumption: a sizeable negative impact of more older workers on [c] is likely to adversely affect labour demand 6

7 Why firm level? - In modern economies, productivity is in essence a firmlevel phenomenon - Individual workers productivity is hardly ever observed. By contrast, many datasets contain good-quality information about firms - production (e.g. sales, value added) - wage/labour costs - workers characteristics (e.g. education; age, gender, labour contract type...) that can aggregated at the firm level and introduced into firm-level equations - Productivity can be examined at macro level but i) aggregation bias exist, + ii) many unobservables 7

8 Age and cognition 8

9 3. Existing firm-level emp. literature Skirbekk, (2008) [International survey]: The most common finding from these studies is a hump-shaped relation between job performance and age,which contrasts almost life-long increases in wages => ratio goes down with age Hellerstein et al. (2007) [USA]: relative productivity of 55+ is.87 whereas rel. wage is 1.128, significant negative effect of age on the ratio 9

10 Aubert & Crépont (2003), [France] productivity rises with age until around the age of 40, before stabilizing, a path which is very similar to those of wages. Productivity/wage ratio deteriorates only for workers aged more than 55 van Ours, J.C & Stoeldraijer, L. (2011), [Netherlands] and Cardoso, Guimaraes & Varejao (2011) [Portugal] find little evidence of an age-related impact on the ratio Cataldi, Kampelmann & Rycx (2011), [Belgium] strong negative effect of older workers on the productivity-labour cost ratio 10

11 Labour productivity 4. Methodology [1] ln (Y i,t /L i,t )= lna + α ln QL i,t +ß lnk i,t lnl i,t where: Y i,t is the firm s value added and QL it a labour quality index à-la-hellerstein-neumark [2] QL it = j µ i,j L i,j,t = µ i,ref L i,t + j ref (µ i,j - µ i,ref ) L i,j,t µ j reflecting the (relative) productivity of type j workers (e.g. prime age workers) 11

12 Assuming same marginal product across firms, (no subscript i) and taking logs QL becomes: [3] ln QL it = ln µ ref + lnl it + ln (1+ j ref (λ j - 1) P ijt ) where λ j µ j /µ ref is the relative productivity of type j workers P ijt L ijt /L it the proportion of type j workers (eg.young, prime age (ref.), old ) 12

13 Since ln(1+x) x, we can linearize [4] Ln QL it = ln µ ref + ln L i,t + j ref (λ j - 1) P i,j,t And the production function becomes: [5] ln(y i,t /L i,t )= lna+α [ln µ ref + ln L i,t + j ref (λ j -1) P ijt ] + ß lnk it lnl ik or, equivalently [6] ln (Y it /L it )= B + (α-1)l it + j ref η j P ijt + ß k it where: B=lnA+α ln µ ref ; P ijt L ijt /L it proportion of type j workers η j = α (λ j 1); λ j =µ j /µ ref j ref. l it =lnl it ; k it =lnk it 13

14 Labour costs [7] W it /L it = j π j L ijt / L it =π ref + j ref. (π j - π ref ) L ijt / L it Taking the logarithm and using log(1+x) x, we can approximate this by: [8] ln(w t /L it )= B w + k ref η W j P i,j,t where B w =ln π 0 η W j = (Φ j 1) Φ j π j / π ref j ref 14

15 Econometric specification The hyp. test = easily implemented if one adopts strictly equivalent econometric specifications [9] ln(y it /L it )= B + (α-1)l it + + j ref η j P itj + ß k it +γf it + ε it [10] ln(w it /L it )= B w + (α w -1)l it + j ref η w j P itj + ß w k it +γ w F it + ε w it Taking the difference of logs <=> log of ratio [11] ln(y it /L it )- ln(w it /L it )= ln(y it /W it )=B G + α G l it + j ref η G j P itj + ß G k it + γ G F it + ε G it where: B G = B -B w ; α G =α-α w ;η G j= η j -η w j; ß G = ß- ß w ;γ G = γ-γ w and ε G it =ε it -ε w it η G j = direct estimate of null hypothesis of no impact on the productivity-labour cost ratio 15

16 The identification problem ln (Y it /L it )= f(l it, k it, P jit ).+ ε it with ε it = ϴ i + τ it + σ it ϴ i unobservable (time-invariant) heterogeneity between firms τ it short-term (asymmetrically) observed productivity shocks σ it random error E(σ it ) = 0 16

17 One can deal with θ i by resorting to first differences or firm fixed effects The biggest challenge= coping with τ it => two methods: IV: lagged values P i,j,t-1 ; P i,j,t-2 as instruments (Aubert and Crépon, 2003, 2007; van Ours & Stoeldraijer, 2011) Intermediates-as-proxy/more structural approach Olley & Pakes (1998), Levinsohn & Petrin (2003), Ackerman, Caves and Fraezer (2006). 17

18 ACF uses intermediates/materials to proxy the short-term productivity term int it =f(τ it, k it, ql it ) Assuming this function is monotonic, it can be inverted Leading to τ it =f -1 (int it, k it, ql it ) ln (Y it /L it )= B+ φ ql it + ß k it + γf it + θ i + f -1 (int it, k it, ql it ) + σ it => Our specificity it to combine this strategy with fixed effects to eliminate θ i => FE-ACF 18

19 4. Data Employers-employees matched data ~ firms with 20+ workers (BELFIRST- BNB) using firm identifiers, we are able to inject information from banque Carrefour de la sécurité sociale/kruispuntbank van de Sociale Zekerheid on the age of (all) workers employed by these firms: ~ workers Data aggregated at firm level Long Panel (9 years) 19

20 Information on firms from the (now dominant) service sector, where administrative and intellectual work is predominant Like Aubert & Crépon (2003) and Dostie (2006), we have a measure of firms productivity (the net valued added), which is measured independently from firms wage cost Contrary to Dostie (2006), we do have a measure of firms capital stock, such that no imputation method is required. 20

21 21

22 22

23 5. Results 23

24 24

25 25

26 Robustness checks 26

27 Conclusion & policy implications Ageing will affect more than welfare systems, as it will also affect the age structure of the workforce. The share of older workers (aged 50+) will rise significantly due to demographics & policy 27

28 Optimists may believe that an ageing workforce will have only a minimal impact on firms performance & labour markets. We produce evidence suggesting the opposite We show that the age structure of firms is a key determinant of their productivity-labour cost ratio (=gross profit margin) Employing larger shares of older workers aged translates into lower profitability ceteris paribus 28

29 Using prime-age workers as a ref., an increase of 10%-points in the share of older workers (50-64) causes a change of productivity of -2.2 to -2.7 % And productivity handicap is not compensated by lower relative labour costs for employers, resulting in a -1.3 to -2.8 % change of the productivity-labour cost ratio (or gross profit margin) 29

30 Policy implications. Need for interventions aimed at - combating the decline of productivity with age (invest in ergonomy, training ) - and/or better adapting labour costs to ageproductivity profiles (reduce the importance of seniority in wage setting, selectively reduce tax wedge beyond 55/60.) 30

31 References Aubert. P. and B. Crépon (2003). La productivité des salariés âgés : une tentative d estimation. Economie et Statistique Cardoso, A., P. Guimarães and J. Varejão (2011), Are Older Workers Worthy of Their Pay? An Empirical Investigation of Age- Productivity and Age-Wage Nexuses, De Economist, 159(2), pp Cataldi, A., S. Kampelmann and F. Rycx (2011), Productivity-Wage Gaps Among Age Groups: Does the ICT Environment Matter?, De Economist, 159(2), pp Levinsohn. J. and A. Petrin (2003). Estimating production functions using inputs to control for unobservables. Review of Economic Studies. 70 (2) Skirbekk, V. (2004), Age and individual productivity: a literature survey, In: Feichtinger, G. (Editor): Vienna yearbook of population research Vienna: Austrian Academy of Sciences Press, pp Skirbekk, V. (2008), Age and productivity capacity: Descriptions, causes and policy options, Ageing Horizons, 8, pp

32 Vandenberghe, V. (2011), Boosting the employment rate of older men and women. An empirical assessment using Belgian firm-level data on productivity and labour costs (2011), De Economist, 159(2), pp Vandenberghe, V. (2013), Are firms willing to employ a greying and feminizing workforce?, Labour Economics, 22, pp Vandenberghe, V. Rigo, M. & Waltenberg, F. (2012), Ageing and Employability. Evidence from Belgian Firm-Level Data, Journal of Productivity Analysis, 40(1), pp Vandenberghe, V. & Lebedinsky, L. (2013), Assessing education's contribution to productivity using firm-level evidence, International Journal of Manpower, 35(8), pp van Ours, J.C & Stoeldraijer, L. (2010), Age, Wage and Productivity, IZA Discussion Papers 4765, Institute for the Study of Labor (IZA), Bonn. 32

33 Appendix

34 A.1 Econometric results overview Key results 34

35 A.2. Employment contract à-la-lazear and extension of the legal age of retirment Current retirement age (55) New retirement age (65) Average prod/l. cost ratio 1 A Actual time-varying ratio B A B Firm 1 Firm 2 Age/seniority 1 Age/seniority 2 A: low wage/high productivity period B: high wage/low productivity perio 35

36 A.3 Weakness of old employment rate in Belgium Women Men 36

37 37

38 A.4 Training Source: Eurostat,

39 A.5. Firm-level evidence on wage profiles and employment (Belgium) 39

40 40

41 A.5. Imperfect substitutability, non-linear specification Vandenberghe, V & A. Ariu (2014), Assessing the role of ageing, feminising and bettereducated workforces on TFP growth, - NBB working paper series, No IRES working paper No

42 Consider a labour-augmented Cobb-Douglas technology Y it =A it K itα (QL it ) β [1] with Q it a labour-quality index à la Hellerstein Neumark (HN), specified as a CES QL it = [µ 1 (L it1 ) ρ +.+ µ n (L itn ) ρ ] 1/ρ [2] L itj, j=1 n labour types (e.g age, gender, blue-vs white collar categories) µ j reflects the (relative) marginal productivity of type j labour ρ the CES substitutability parameter A it =A i0 e τ.t + ω it; with A i0 the starting value of firm i s TFP, e τ.t + ω it capturing its dynamic τ is the common annual rate of growth; ω it the firm-specifc term, with ω it =ϴ i + δ it containing a fixed effect 42

43 43

44 AGE 44