Upload
liang
View
21
Download
0
Tags:
Embed Size (px)
DESCRIPTION
Smarter task assignment or greater effort: what makes a difference in team performance? Burgess, Propper, Ratto, Scholder, Tominey. Introduction. Two main motivations: Are financial incentives useful in the public sector? Do they have a role to play in public service reform? - PowerPoint PPT Presentation
Citation preview
Policy Studies Institute seminar, 26 June 2007
1
Smarter task assignment or greater effort: what makes a difference in team performance?
Burgess, Propper, Ratto, Scholder, Tominey
Policy Studies Institute seminar, 26 June 2007
2
Introduction
Two main motivations: Are financial incentives useful in the
public sector? Do they have a role to play in public service reform?
Compared to theoretical work, relatively little empirical evidence on incentives in organisations, and how they work.
Policy Studies Institute seminar, 26 June 2007
3
What we do:
Study a pilot scheme of team-based financial incentives in HM Customs & Excise (as was)
Use rich and disaggregate dataset based on their personnel records and MIS
Research design allows diff-in-diff-in-diff, and allows us to understand in detail how the scheme worked.
Policy Studies Institute seminar, 26 June 2007
4
What we find: At team level the incentive scheme did
increase performance The incentive structure did raise individual
productivity. Managers reallocated efficient workers to
the incentivised tasks; this happened disproportionately in one of the treatment teams.
This reallocation was the more important contributor to the overall outcome.
Why? Maybe career concerns …
Policy Studies Institute seminar, 26 June 2007
5
Plan
(Brief) details of the scheme Data Results:
Team level Worker productivity Reallocation Decomposition
Conclusions – career concerns?
Policy Studies Institute seminar, 26 June 2007
6
UK experience on PRP in the public sector November 1998 Public Services Productivity Panel. Advise the government of ways of improving the productivity and efficiency of government departments.
Programme to improve efficiency and productivity in the public sector. Makinson report (2000). Some directions and particular emphasis on teamwork.
Team-based pilot schemes in DWP and HM Customs and Excise.
Policy Studies Institute seminar, 26 June 2007
7
(Brief) details of the scheme 1 Pilot ran 1 April – 31 Dec 2002, with 12
trials Team-based: team = division, a few
offices. Bonus for team as a whole. Pairs of trial teams, plus blind control,
with different bonus schemes. We use pair of VAT Assurance
divisions, team 1 (154 workers); team 2 (158); and control team (281)
Mean potential bonus about 4% salary.
Policy Studies Institute seminar, 26 June 2007
8
Team 1 Team 2 Jobband 2 420 740 Jobband 3 515 740 Jobband 4 550 740 Jobband 5 453* 740 Jobband 6 720 740 Jobband 7 860 740 Jobband 8 970 740 Jobband 9 1145 740 Jobband 11 1520 740 * only observation, however, the person did not work all 275 days, so did not get the total bonus..
Bonus amount
Policy Studies Institute seminar, 26 June 2007
9
(Brief) details of the scheme 2 Each incentive target was baseline + 5% Aware of multi-task issues, and controlled. Some tasks incentivised, some not. Output split into different activities, different
client groups and different outputs 5 targets: 2 on visits (time), 3 on yield (££) Note different production function effects of
time and yield.
Policy Studies Institute seminar, 26 June 2007
10
Activities of VAT assurance office
Trader groups No. of audits (time)
Tax revenue (yield)
Non-trader work - Other trader work - None-core - Unregisterable entity - Trader audit work divided into the following trader groups: Non-VAT - - Non-trader audit - - New registration Low risk - - Medium risk - - High risk - - Exceptional risk Large traders - Corporate groups - - Branches - - Insolvent - - Deregistered - - Missing traders - -
Policy Studies Institute seminar, 26 June 2007
11
Data
9 months of the scheme, same 9 months before
We have records for each worker, in each week, and in each trader group.
Outcomes: time allocated by each worker to each trader group (no direct measure of visits), total yield per worker per trader group, positive and negative yield per worker per trader group, and productivity.
Focus on continuously employed, front-line staff
1 Apr 2001
31 Dec 2001
1 Apr 2002
31 Dec 2002
Policy Studies Institute seminar, 26 June 2007
12
Data 2 Time:
Trader audit is more than half of time Mean time per worker/week/group is 8 hrs Mean groups worked on per w/w is 3 Each incentivised group is about 30% of time
Yield: Mean +ve yield per w/w/g is £302k; but median
is just £5k outliers! One obs w/w/g of £280m. Largest share of +ve yield from exceptional
risk TG Av. –ve yield much smaller, follows same
distribution
Policy Studies Institute seminar, 26 June 2007
13
Results
Team level Worker productivity Strategic task allocation Decomposition
Policy Studies Institute seminar, 26 June 2007
14
Results - Team level
Exploit variation in incentivisation across: Time Treatment status Trader group
Diff-in-diff-in-diff, controlling for time effects, team effects and TG effects.
Use median for anything involving ££ Short answer: scheme worked in T2
more than T1.
Policy Studies Institute seminar, 26 June 2007
15
Table 1a Table 1b Table 1c For all trader groups For all incentivised For all non-incentivised Differences between incentivised & pre-incentivised period
Differences between incentivised & pre-incentivised period
Differences between incentivised & pre-incentivised period
Variable Team
1 Team 2 C Variable Team
1 Team 2 C Variable Team
1 Team
2 C
Total Time 50.5 103.35 41.5 Total Time -3.8 73.89 -1.5 Total Time 13.5 -13.4 1.099
Total Yield 67193 73441.5 0 Total Yield 10901 31725 0 Total Yield 6406 0 0 Positive Yield 84983 80877 0
Positive Yield 8370 46550.5 0
Positive Yield 6406 0 0
Negative Yield 0 0 0
Negative Yield 0 0 0
Negative Yield 0 0 0
Productivity 137.54 120.13 0 Productivity 58.70 89.78 9.17 Productivity 58.24 0 0
Table 1d Table 1e Table 1f Difference-in-Difference
Difference-in-Difference
Difference-in-Difference
Variable T1-C T2-C Variable T1-C T2-C Variable T1-C T2-C
Total Time 9 61.85 Total Time -2.3 75.40 Total Time 12.4 -14.5
Total Yield 67193 73441.5 Total Yield 10901 31725 Total Yield 6406 0 Positive Yield 84983 80877
Positive Yield 8370 46550.5
Positive Yield 6406 0
Negative Yield 0 0
Negative Yield 0 0
Negative Yield 0 0
Productivity 137.54 120.13 Productivity 49.53 80.61 Productivity 58.24 0
Variable M*(T1-TC) – nonM*(T1-TC) M*(T2-TC) – nonM*(T2-TC)
Total Time -14.7 89.89
Total Yield 4495 31725
Positive Yield 1964 46550.5
Negative Yield 0 0
Productivity -8.71 80.61
Difference:
Diff in diff:
Diff in diff in diff:
Policy Studies Institute seminar, 26 June 2007
16
Analysis – Worker level (unit: worker-trader group) Same five outcomes available at w/g level,
adding up over 9 months, and differencing Regress:
Specifications: With & w/out non-VAT; with & w/out
trimming Treated teams combined (intercept
dummies), and each treated team + control separately
gigiiig Incentivey γ.GTeamδXβ ...
Policy Studies Institute seminar, 26 June 2007
17
Policy Studies Institute seminar, 26 June 2007
18
Worker Effort and yield
These are the coefficients on incentivisation
OLS on difference in outcome (so implicitly FE).
Unit is worker/group per 9-month spell, in £
This specification is trimmed of outliers (top & bottom 5%), and excludes non-VAT outcomes.
Other controls: TG effects, team effects, years in grade, age, mean age squared, gender, part-time/full-time worker, job band.
Policy Studies Institute seminar, 26 June 2007
19
At team level, treated teams spend more time on all trader groups and recover more yield than the control team.
At individual level, the increase in productivity in incentivised trader groups is larger in Team 1 than in Team 2, and statistically significant only for Team 1.
Mismatch between individual-level and team-level outcomes: individual productivity increases more in T1 than in
T2. at team level T2 performs better than T1
Summary of results
Different strategy at team level between the treated teams
Policy Studies Institute seminar, 26 June 2007
20
Results – Task Allocation
Identify efficient workers in pre-scheme period as those in top quartile of productivity
Consider reallocation across a four way split: incentivised vs non-incentivised tasks, efficient vs non-efficient workers.
Is there a significant difference in the reallocation of time on the incentivised and the non-incentivised trader groups by efficient workers and other workers in each team?
Policy Studies Institute seminar, 26 June 2007
21
Inc
Non-inc
Time allocation
Increasing
total time
Reallocating fixed time
Horizontal axis = time on incentivised trader groups
Vertical axis = time on non-incentivised trader groups
Teams can increase total time on audit, and/or reallocate time between the two types of groups
Policy Studies Institute seminar, 26 June 2007
22
Inc
Non-inc
Time allocation
Any point (x, y) derives from:
- an overall increase in time of (x – y)/2, and
- a net reallocation of (x + y)/2.
- The latter is half of the vertical distance between (x, y) and the positively sloped 45o line.
-The gross reallocation is twice that, namely the full vertical distance between (x, y) and the positively sloped 45o line
x, y
Policy Studies Institute seminar, 26 June 2007
23
Strategic task allocation: efficient workers
-122
Change in time spent on NI_TG
T2
87
T1
49
-89C
29
Change in time spent on I_TG276
Policy Studies Institute seminar, 26 June 2007
24
All teams Team 1 Team 2 Control team Efficient workers 131.94 87.12 275.91 48.51 Others 56.89 35.02 98.84 45.92
t = 2.98 t = 1.50 t = 3.46 t = 0.06
Significant at 1% Significant at 10% Significant at 1% Not significant
All teams Team 1 Team 2 Control team Efficient workers -63.56 29.31 -122.17 -89.29 Others 26.04 54.05 -8.31 28.73
t = -4.30 t = -0.741 t = -2.66 t = -3.75
Significant at 1% Not significant Significant at 1% Significant at 1%
Individual level difference in time spent per month Incentivis
ed trader groupsNon-incentivised trader groups
Reallocation between incentivised and non-incentivised trader groups All teams Team 1 Team 2 Control team
Efficient workers 195.50 57.80 398.08 137.79 Others 30.84 -19.03 107.15 17.18
t = 4.56 t = 1.52 t = 3.91 t = 2.1386
Significant at 1% Significant at 10% Significant at 1% Significant at 5%
Efficient workers
Team 1 vs. Team 2 Team 1 vs. Team C Team 2 vs. Team C Team X 57.80 57.80 398.08 Team Y 398.08 137.79 137.79
t = -4.25 t = -1.13 t = 2.92
Significant at 1% Not significant Significant at 1%
Reallocation of efficient workers
Team 1 vs. Team 2 Team 1 vs. Team C Team 2 vs. Team C Team X -19.03 -19.03 107.15 Team Y 107.15 17.18 17.18
t = -2.96 t = -1.0241 t = 2.21
Significant at 1% Not significant Significant at 5%
Reallocation of other workers
Policy Studies Institute seminar, 26 June 2007
25
Team 1 Team 2
Efficient worker * New registrations -15.001 (21.992)
43.716 * (25.351)
Efficient worker * Low risk -3.071 (8.022)
-26.502 *** (9.769)
Efficient worker * Median risk 14.049 (9.806)
-18.213 ** (8.882)
Efficient worker * High risk -15.169 (20.790)
-45.998 (29.948)
Efficient worker * Exceptional risk 62.801 ** (25.398)
122.785 *** (42.559)
Efficient worker * (999) Large traders -14.579 (23.095)
-55.919 (41.144)
Efficient worker * Corporate groups 0 (0.000)
-9.942 (6.982)
Efficient worker * Branches 0.495 (4.063)
0 (0.000)
Efficient worker * Insolvent 9.547 ** (3.865)
4.715 (3.299)
Efficient worker * Deregistered 3.002 (3.116)
9.241 ** (3.895)
Efficient worker * Missing traders 10.339 (6.056)
-0.232 (2.515)
Time change per trader group
Policy Studies Institute seminar, 26 June 2007
26
Results - Decomposition
How much of the change in a team’s yield is due to increased worker effort, and how much to better deployment of workers?
Tijt is time spent by i on group j period t, and yijt is the corresponding productivity
Decomposition of total team yield: 21211212 ijj
ijiji
ijj
ijiji
TTTYYY
Policy Studies Institute seminar, 26 June 2007
27
Change in yield due to: Team 1 Team 2 Control Team
Reallocation 10,500,514 19,785,364 4,330,667 Increased Effort 3,445,083 1,378,579 752,514 Total 13,945,597 21,163,943 5,083,181 Reallocation as % 75.3 % 93.5 % 85.2 % Increased Effort as % 24.7 % 6.5 % 14.8 %
Decomposition of the change in total yield
We have taken out the highest single yield datapoint in T2 of £280m
Policy Studies Institute seminar, 26 June 2007
28
Conclusions
The incentive scheme: raised individual yield and productivity led to the reallocation of efficient workers
towards the incentivised tasks. Reallocation was the more important contributor.
Comparing the teams: The increase in individual yield collected was
essentially the same in the two teams. But T2 engaged in reallocation to a significantly
greater degree than T1 T2 hit its targets and collected the bonus. Why?
Policy Studies Institute seminar, 26 June 2007
29
Conclusions 2 Why: Staff composition: more experience in T2? No.
Different managerial strategy prior the scheme? No.
Nature of bonus scheme? No. Intrinsic incentives – career concerns?
Office managers in T2 were younger, and had been in that job band for fewer years.
Division manager in T2 was much younger than in either the control or T1, and had less experience.
Managers in T2 may have been more driven to perform well for longer term career concerns
Policy Studies Institute seminar, 26 June 2007
30
Policy Studies Institute seminar, 26 June 2007
31
Job band 9 Job band 11
All Team 1 Team 2 Control All Team 1 Team 2 Control
gender 0.24 0.10 0.33 0.27 0.67 1.00 1.00 0.00age 50.62 52.90 46.78 51.40 48.00 48.00 40.00 56.00job_band 9.00 9.00 9.00 9.00 11.00 11.00 11.00 11.00jobband_yrs 4.09 4.60 2.78 4.53 4.33 5.00 2.00 6.00service 27.94 27.70 25.44 29.60 28.00 27.00 21.00 36.00pay 30970.32 30829.90 29558.22 31911.20 39651.00 38667.00 38279.00 42007.00gmpay 2526.78 2485.22 2449.04 2601.13 3304.25 3222.25 3189.92 3500.58gmpot 2526.78 2485.22 2449.04 2601.13 3304.25 3222.25 3189.92 3500.58otm 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00pt 0.12 0.10 0.11 0.13 0.00 0.00 0.00 0.00ptpw 28.33 24.00 . 30.50 . . . .sick 0.03 0.00 0.11 0.00 0.00 0.00 0.00 0.00days 271.61 268.90 275.00 . 275.00 275.00 275.00 .bonus 915.95 1079.40 734.33 . 1130.00 1520.00 740.00 .
Managers Profile
Policy Studies Institute seminar, 26 June 2007
32
The Teams
There are 12 teams, consisting of trial sites in RBS and LBG. These are:
- VAT Assurance Division, Solent, RBS South
- VAT Assurance Division, North East, RBS Central
- Excise Assurance, RBS Scotland
- Excise Assurance, RBS North
- International Trade Assurance, RBS North
- International Trade SME Assurance, RBS Scotland
- Risk Team (Operational Analysis Division), RBS North
- Risk Team, RBS Northern Ireland
- National Function – Registration Unit
- National Function – Debt Management Unit
- National Function - Nation Advice Service
- LBG Scotland and Northern Ireland
Policy Studies Institute seminar, 26 June 2007
33
Are the obligations/targets in all other areas of the team activity which are not subject of the incentives targets met?
No
Yes
The team qualifies for admission to bonus scheme
Does the team meet the baseline target and at least 50% of the stretch on all incentive targets?
No
Yes
Payment of bonus:
- 100% bonus if all the incentive targets are met in full- 50% bonus if incentive targets are not met in full but baseline target and at least 50% of the stretch is met on all incentive targets
Bonus award process
Policy Studies Institute seminar, 26 June 2007
34
All staff Continuous frontline staff
Overall Team 1 Team 2 Control Overall Team 1 Team 2 Control
Gender 0.47 0.39 0.44 0.52 0.44 0.40 0.40 0.50
Age 44.2 45.7 45.4 42.8 44.6 45.0 45.5 43.7
Job band 5.9 6.2 5.9 5.8 6.2 6.3 6.3 6.0
Years in job band 3.9 4.0 4.3 3.6 4.2 4.0 4.7 4.1
Years at Customs &Excise 18.1 19.1 19.1 17.1 19.5 19.3 20.0 19.2
Annual pay (£) 20247 20797 20066 20047 20671 20742 20789 20551
Gross monthly pay (£) 1594 1630 1599 1571 1632 1619 1657 1625
Gross monthly pay + overtime (£) 1599 1632 1599 1582 1639 1621 1657 1639
Overtime per month (hours) 0.38 0.16 0.02 0.70 0.44 0.15 0.02 0.88
Part-time marker 0.16 0.16 0.13 0.17 0.16 0.19 0.12 0.16
Part-time hours per week 24.8 24.1 - 25.1 24.7 23.8 - 25.2
Days of sickness 1.1 1.1 1.5 0.9 1.0 1.2 1.3 0.8
Days in scheme 260.2 258.8 261.5 - 264.0 262.4 265.6 -
Potential bonus (£ per person) - 688 681 - - 695 696 -
Team staff mean characteristics.
Policy Studies Institute seminar, 26 June 2007
35
400
06
000
800
01
0000
120
001
4000
2001m1 2001m7 2002m1 2002m7 2003m1Month
Trial Team 1 Trial Team 2Control Team
Total monthly time (hours)
Time series variation in time.