Presentation on theme: "Centre for Market and Public Organisation How important is pro-social behaviour in the delivery of public services? Paul Gregg, Paul Grout, Anita Ratcliffe."— Presentation transcript:
Centre for Market and Public Organisation How important is pro-social behaviour in the delivery of public services? Paul Gregg, Paul Grout, Anita Ratcliffe Sarah Smith and Frank Windmeijer University of Bristol, CMPO
Overview Growing literature emphasising importance of intrinsic motivation for workers in the non-profit sector Is there evidence that workers in the non-profit sector (public and not- for-profit sectors) provide more effort in the form of donated labour than workers in private sector? Does the sector affect behaviour? Analysis of British Household Panel Survey on prevalence of doing unpaid overtime
Policy background Importance of contracting out of public services to private and not-for-profit providers Possibility of pro-social behaviour unique to non-profit provision is an important dimension in the debate about who should provide public services Total spending on services Procurement of services….. Of which Frontline service delivery … Of which Facilities (catering, cleaning) Health£88.7 bn£20.7 bn43%41% Education£61.1 bn£5.4 bn46%24% Local govt services£24.9 bn£4.9 bn83%4% Source: Oxford Economics, 2008
Growing literature emphasizing that workers in the non-profit sector (public sector and not- for-profit sector) may be intrinsically motivated to work Common emphasis on association between sector of employment and pro-social behaviour; differences in mechanism through which this arises Institutional form (Francois, 2000) –Intrinsically motivated individuals will choose not to donate their labour in a for-profit firm because the firm will respond by reducing other inputs in order to increase profit –In a not-for-profit firm, the non-distribution constraint means that extra effort benefits service users. In the public sector, bureaucratic budget-setting has the same effect –Institution matters: individuals donate labour in a non-profit firm, not in a for-profit firm Mission-matching (Besley and Ghatak, 2005) –Individuals differ in the degree to which they care (mission) –They will donate labour in whatever sector they work in, but will be attracted to perceived caring sectors (=non-profit)
Empirical research Evidence that public service motivation is reflected in what people say: Interviews reported in the Guardian: Does motivation vary by sector? –Children's service manager, NFP: I couldn't do it for profit –Children's strategy manager, Public: I wouldn't move into the private sector, because at the end of the day it's a business –Housing worker, Public: I prefer to be with the state sector because you can do more to help people. With the private sector.. there's more of a profit angle British Social Attitudes Survey (John and Johnson, 2008): –Public sector employees are twice as likely as private sector employees to say that it is very important to them that a job is useful to society (32% compared to 15%) –More likely to say that it is very important that a job allows them to help other people (27% compared with 19%)
Possible halo effect? Important to look at what people do, not just what they say Rotolo and Wilson (2005) – civic-mindedness of public sector workers highlighted by greater propensity to volunteer Frank and Lewis (2004) – higher level of (self-reported) effort in public sector, but public sector defined by industry Mokan and Tekin (2005) – detailed study of childcare industry using rich employer- employee matched data set. Whether people value an important job has a negative and significant effect on wages only in the non-profit sector.
Our contribution Do people working in the non-profit sector donate more labour than people working in the for-profit sector? –Estimate probability of doing unpaid overtime by sector of employment including numerous controls for individual and job characteristics If so, what is the likely mechanism – institution or selection? –Estimate fixed effects model to see what happens to the probability of doing unpaid overtime when people change sector –Look at behaviour of switchers
Data British Household Panel Survey Followed > 10,000 individuals each year since 1991 Our sample covers 1993-2000 (matched with wage information at the occupation level from the Labour Force Survey) Select full-time employees (30+ hours). No differential selection by sector. 24135 obs aged 16-60; 6,601 individuals
Some definitions Measure of donated labour – unpaid overtime –Thinking about your (main) job, how many hours excluding overtime and meal breaks are you expected to work in a normal week? –And how many hours overtime do you usually work in a normal week? –How much of that overtime (usually worked) is usually paid overtime? –27% individuals do unpaid overtime in BHPS; 29% in LFS
Some definitions Sector of employment –Which of the types of organisations do you work for (in your main job)? –For-profit – private firm/ company –Non-profit – civil servant/central government, local government/town hall, NHS or higher education, nationalised industry, non-profit organisation –No significant differences in behaviour between individuals working in the public and not-for-profit sectors, although nfp sector is small
Some definitions Caring services based on industry classification (SIC1980) –Caring – health, education, social care (=17% sample) –Non-caring – all other industries Results robust to alternative definitions –Narrower – cross-classifying industry with occupation to include only managers, natural scientists, health & teaching professionals and childcare workers (=14% sample) –Wider – including R&D, the arts & culture corresponding to industries where NfPs are located according to Rose-Ackerman, 1996 (=20% sample)
Full sampleHealthEducationSocial care Non-profit caring14.80%80.04%88.60%83.91% For-profit caring2.70%19.96%11.40%16.09% Non-profit non-caring13.34% For-profit non-caring69.16% Total2413514731825926 Non-profit refers to not-for-profit organisations and public organisations For-profit refers to private firms Caring refers to health, education and social care Non-caring refers to all other industries Distribution by sector
Unpaid overtime, by sector of employment Proportion of sample Proportion doing unpaid overtime Mean hours unpaid OT (>0) Caring services, non-profit0.1480.469.59 Caring services, for-profit0.0270.29*8.34* Other industries, non-profit0.1330.22*6.56* Other industries, for-profit0.6920.24*8.49* Full-time employees; caring services = health, education and social care * Indicates that the difference to caring services, non-profit is significantly different at 5% level
Proportion doing unpaid overtime, by sector of employment Public sectorNot-for-profit All Caring services0.460.49 Non-profitFor-profit Health0.310.23* Education0.590.47* Social care0.430.15* Women0.460.27* Men0.460.34*
Pooled regression model D it = dummy variable if individual i, i=1,…,N, does any unpaid overtime in time t Sector it = set of four binary indicators representing the non-profit and for-profit caring sectors and the non-profit and for-profit non-caring sectors x = vector of individual characteristics: age, age-squared, gender, married, presence of children, age of youngest child (interacted with gender), education, ethnicity z = vector of job characteristics: wage measures, contracted hours, trade union present (and whether the individual is a member), pension scheme (and whether the individual is a member), individual is a manager, workplace size, indicators for industry (health, education and social care) Region and year dummies
Career concerns Individuals motivated to do unpaid overtime by the prospect of higher future remuneration, i.e. career concerns Bell and Freeman (2001) show that hours worked are positively related to occupational wage dispersion (measured by standard deviation of ln hourly earnings) We take a similar approach but look at age-relevant part of wage distribution, i.e. 16-60 for individuals aged 16-30, 30-60 for individuals aged 30-45 and 45-60 for individuals aged 45-60 Both within occupation, across sectors and within occupation, within sectors
Career concerns Since current wage may reflect past unpaid overtime (and be correlated with current unpaid overtime), instrument using ln of median wages by occupation/ year and age group Opportunity cost, income effect and/or career concerns Other variables to capture career concerns: Quadratic in years tenure in current job Indicator: Do you have an opportunity for promotion in your current job? Indicator: Does your pay include a bonus? Indicator: Whether individuals are satisfied with job security
Results for pooled linear probability model Dependent variable = whether individual does unpaid overtime (0/1) (1)(2)(3) For-profit, caring (omitted) Non-profit, caring0.174*** (0.032) 0.139*** (0.032) 0.123*** (0.027) Non-profit, non-caring-0.062** (0.031) -0.053 (0.030) -0.148*** (0.032) For-profit, non-caring-0.045*** (0.029) 0.003 (0.027) -0.118*** (0.030) ln wage occ/age/year0.326*** (0.015) SD ln wage, occ/age/year0.364*** (0.048) Control variablesNo controlsControls for age, educ, gender, marital status, children, ethnicity Additional controls for job tenure, promotion opportunities, bonus pay, job satisfaction, manager, firm size, unionisation, usual hours, industry N (indivs)24135 (6016) Standard errors are clustered at the individual level. *** significant at the 1% level
Hours worked, by sector of employment Proportion of sample Contracted hours + unpaid OT Proportion doing paid overtime Mean hours paid OT (>0) Contracted hours + unpaid OT + paid OT Caring services, non-profit0.14841.440.107.9042.22 Caring services, for-profit0.02740.34*0.22*7.2142.10 Other industries, non-profit0.13339.53*0.26*8.2141.66 Other industries, for-profit0.69241.320.34*8.5144.20* Full-time employees; caring services = health, education and social care * Indicates that the difference to caring services, non-profit is significantly different at 5% level
Results for pooled OLS model Dependent variable = ln total hours Basic hours + unpaid OTBasic hours + unpaid OT + paid OT For-profit, caring (omitted) Non-profit caring0.0234*** (0.011) 0.0003 (0.011) Non-profit, non-caring-0.0633** (0.012) -0.0632*** (0.013) For-profit, non-caring-0.0451*** (0.012) -0.0235 (0.012) Control variablesFull set of individual and job controls
Individuals in the non-profit caring sector are significantly more likely to do unpaid overtime than in the for-profit caring sector Not a general non-profit effect – applies only in caring industries But they do not work more hours in total; they do less paid overtime Is unpaid overtime voluntary donated labour or just a social norm? Use fixed effects regression to look at what happens when people change sector
Fixed effects regression model Table 5. Switches across sectors Sector, time t Sector, time t – 1 Non-profit caring For-profit caring Non-profit noncaring For-profit Noncaring N-P caring24048313550 F-P caring80288588 N-P noncaring12992224184 F-P noncaring888513312099
Results for fixed effects linear probability model Dependent variable = whether individual does unpaid overtime (0/1) (1)(2)(3) For-profit, caring (omitted) Non-profit, caring0.000 (0.029 ) -0.001 (0.028) 0.002 (0.028) Non-profit, non-caring-0.042 (0.030) -0.039 (0.030) -0.061 (0.042) For-profit, non-caring-0.015 (0.027) -0.015 (0.027) -0.037 (0.041) Ln wage occ/age/year0.092*** (0.017) SD ln wage, occ/age/year0.110*** (0.040) Control variablesNo controlsControls for age, educ, gender, marital status, children, ethnicity Additional controls for job tenure, promotion opportunities, bonus pay, job satisfaction, manager, firm size, unionisation, usual hours, industry N (indivs)22703 (4619) Standard errors are clustered at the individual level. *** significant at the 1% level
Results for fixed effects linear probability model Dependent variable = whether individual does unpaid overtime (0/1) (1) For-profit, caring (omitted) First period: For-profit caring-.0247 (.0339) First period: Non-profit caring-.0138 (.0378) Subsequent periods: Non-profit caring.0019 (.0361) First period: Noncaring-.0320 (.0461) Subsequent periods: Noncaring-.0683 (.0477) Control variablesAdditional controls for job tenure, promotion opportunities, bonus pay, job satisfaction, manager, firm size, unionisation, usual hours, industry N (indivs)22703 (4619)
Fixed effects results Insufficient switchers? –Estimated coefficient is (close to) zero, rather than being imprecisely estimated Measurement error? –Would have to be very high (around a half) to generate our observed results –75% of switchers (from n-p care to f-p care or v.v) stay in next sector for at least two periods (i.e. not just one-off mis-reporting) –Other sector coefficients are non-zero
Fixed effects results No change in behaviour on switching –Evidence against social norms since individuals would change to comply with behaviour in new sector –But also inconsistent with strong organisational form explanation (Francois, 2000) individuals donate labour in non-profit, but not in for-profit sector –Is there any evidence to support a selection story?
Evidence on selection Compare the switchers with the stayers For people working in the non-profit caring sector: Do people who (ever) switch from the non-profit sector to the for-profit sector or the non-caring sector donate less labour when they are in the non-profit sector than people who stay? For people working in the for-profit caring sector: Do people who (ever) switch from the for-profit sector to the non-profit sector donate more labour when they are in the for-profit sector than people who stay?
Evidence on selection Estimation results for linear probability model Dependent variable : whether individual does unpaid overtime (0/1) Employees in the non-profit caring sector Employees in the for-profit caring sector Switch to for-profit caring-0.132*-0.114** (0.075) (0.058) Switch to non-profit caring0.0780.039 (0.089) (0.069) Switch to non-caring-0.141***-0.064-0.0530.025 (0.052) (0.044) (0.076) (0.068) Control variablesNoYesNoYes N3134517 Robust standard errors are clustered at the individual level *** indicates significant at 1% level, ** at 5% level, * at 10% level
Conclusions Evidence of an association between institutions and donated labour Institutions appear to work through selection rather than incentives Possible that the behaviour of some people may be affected by the sector that they work in, but we dont observe them switching An exogenous change in institution might be more convincing to identify the effect of sector, although selection would still be important Sample sizes limit the extent to which we can carry out more detailed analysis of switchers What is it about the non-profit sector that attracts pro-socially motivated people?