High-level Knowledge Production

Download Report

Transcript High-level Knowledge Production

HIGH LEVEL KNOWLEDGE PRODUCTION:
ANALYSES OF INPUTS AND OUTPUTS
UCT Strategy Forum
23 March 2012
1
INTRODUCTION
1
In its annual surveys of research and experimental development in SA,
the HSRC follows the OECD in identifying these as higher education’s
key inputs and outputs:
Inputs: doctoral enrolments and academic staff
Outputs: research publications and doctoral graduates
2
This presentation falls into three parts:
•
a discussion of these inputs and outputs at a national level;
•
a discussion of the research incentives built into the current
government funding framework;
•
a brief account of institutional differentiation, using UCT as an example.
2
PART I: OVERVIEW OF NATIONAL INPUTS & OUTPUTS
SUMMARY OF HIGH LEVEL INPUTS & OUTPUTS
Graph 1:
Graph 2:
National totals of high level knowledge inputs & outputs: 1996-2010
Average annual increases in high level knowledge inputs & outputs
DOCTORAL ENROLMENTS
Graph 3:
Graph 4:
Graph 5:
% of doctoral enrolments in each broad field of studies
Doctoral enrolments by gender
Doctoral enrolments by race group
DOCTORAL GRADUATE EFFICIENCY
Graph 6:
Graph 7:
Graph 8:
Graph 9:
Doctoral graduation rates & cohort output equivalents
Progress of the HE system's 2001 cohort of new doctoral students
Graduates and dropouts from combined 2001, 2002, 2003 intakes of new
doctoral students
Comparisons of actual totals of doctoral graduates and normative totals on
National Plan target ratio: 2000-2010
ACADEMIC STAFF
Graph 10:
Graph 11:
Graph 12:
Total permanent academic staff: 1996-2010
% of academic staff with doctorates by institutional category
Research publications per permanent academic by institutional category
3
SUMMARY OF HIGH LEVEL INPUTS & OUTPUTS
3 Graph 1 offers summaries for the 15-year period 1996-2010. Doctoral
enrolments were 1.3% of national total of 893 000 students in 2010.
Graph 1:
National totals of high level knowledge inputs & outputs:
1996-2010
18000
16000
14000
13449
14184
14673
15809
15423
15936
13098
12000
11468
10000
8790
8000
6000
4000
5622
5528
5164
5456
685
761
6394
5936
7763
6483
6660
969
1104
9800
9939
8003
8353
1100
1182
2000
0
16684
1996
Doctoral enrolments
1998
961
2000
Doctoral graduates
2002
2004
2006
Research publications
2008
9748
1421
2010
Permanent academics
4
4 Graph 2 divides growth rates between (a) 1996 and 2002, which covered the
period of the 1997 HE White Paper and the 2001 National Plans, and (b) 2004-2010
which covered the implementation of the new 2003 funding framework.
Graph 2: average annual increases in high level knowledge inputs &
outputs
8.0%
7.0%
7.0%
6.6%
6.0%
5.9%
6.0%
5.4%
4.5% 4.3%
5.0%
4.0%
4.0%
2.4%
3.0%
1.5%
2.0%
1.6%
1.3%
1.0%
0.0%
1996-2002
Doctoral enrolments
Doctoral graduates
2004-2010
Research publications
1996-2010
Permanent academics
5
DOCTORAL ENROLMENTS
5 Next three graphs offer further information on national doctoral enrolments
Graph 3: % of doctoral enrolments in each broad field of studies
60%
50%
40%
50%
45%
41%
45%
42%
37%
44%
40%
43%
40%
45%
38%
47%
36%
30%
49%
33%
20%
10%
0%
11%
10%
3%
3%
1996
1998
10%
3%
2000
Science and technology
10%
6%
2002
11%
7%
2004
Business & management
11%
6%
2006
Education
10%
7%
2008
10%
8%
2010
Humanities
6
Graph 4: doctoral enrolments by gender
70%
66%
60%
64%
64%
62%
60%
58%
57%
58%
40%
42%
43%
42%
50%
40%
34%
36%
38%
36%
30%
20%
10%
0%
1996
1998
2000
2002
Female
2004
2006
2008
2010
Male
6 Graph 5 on the next slide shows how the race group composition of doctoral
enrolments has changed over 1996-2012. Some of the change results from the
recruitment of international students.
7
7 In 2010 2840 (or 56%) of the African total consisted of African students from
other African countries.
Graph 5: doctoral enrolments by race group
6000
5066
5000
4861
4819
4568
4853
4486
4000
4020
3875
4022
3993
3583
3000
2933
2239
2000
1610
1000
0
1053
344
256
683
264
197
1996
1998
African
464
327
2000
Coloured
619
419
2002
768
529
2004
Indian
813
585
2006
774
868
575
681
2008
2010
White
8
DOCTORAL GRADUATE EFFICIENCY
8 Graph 6 offers a first picture of the doctoral output efficiency of SA’s
universities, based on the output ratios of the 2001 National Plan. The National
Plan set this as an output norm:
these
The ratio between doctoral graduates in any given year and doctoral
enrolments in that same year should = 20%. So if 10 000 doctoral
students were enrolled in the HE system in year X, then at least 200 of
students should graduate in year X.
9 This norm was based on a further target norm that at least 75% of any cohort of
students entering doctoral studies for the first time in (say) year Y, should
eventually graduate. If the cohort output norm is to be achieved, then the 20%
ratio of total graduates to total enrolments would have to be met over a period of
time.
10 Graph 6 shows that, as far as doctoral outputs are concerned, the public HE
system has failed to meet the National Plan’s efficiency targets. The graph
suggests that less than 50% of students who entered doctoral programmes in SA
between 2000 and 2010 will eventually graduate.
9
Graph 6: doctoral graduation rates & cohort output
equivalents
80%
75%
70%
60%
52%
50%
45%
45%
40%
30%
20%
20%
14%
12%
12%
10%
0%
1998-2002
2002-2006
Ratio of graduates to enrolments
2006-2010
National target
Cohort graduation equivalent
11 The tracking over time of cohorts of new doctoral students confirms that a 50%
graduation rate is what can be expected in SA. Graph 7 summarises the progress
of the 2001 intake of 1722 new doctoral students: 925 have graduated and 797
dropped out.
10
Graph 7: progress of the HE system's 2001 cohort of new
doctoral students
60%
54%
46%
50%
40%
30%
29%
20%
10%
10%
7%
0%
Dropped out:
years 1 or 2
Dropped out:
years 3 to 5
Dropped out:
years 6 or 7
Total drop-outs
% of original
cohort
graduating
11
12 Graph 8 summarises the final status of the new doctoral students enrolling in
either 2001 or 2002 or 2003.
Graph 8: graduates and dropouts from combined 2001, 2002, 2003
intakes of new doctoral students
2767 (46% )
3219
TOTAL
5986
1523 (50% )
1517
White
3040
254 (46% )
301
555
Indian
130 (47% )
144
274
Coloured
860 (41% )
African
1257
2117
0
1000
2000
New intake total
3000
4000
Dropped out
5000
6000
7000
Graduated
12
13 Graph 8 offers estimates of the effects of output inefficiencies in SA’s doctoral
programmes.
Graph 9: comparisons of actual totals of doctoral graduates
and normative totals on National Plan target ratio: 2000-2010
25000
19995
20000
15000
12522
10000
5000
0
-5000
-10000
Actual graduates
produced
National Plan target
Shortfall
-7473
13
ACADEMIC STAFF
14 Academic staff with doctoral degrees are a key input for high level knowledge
production. Permanent academic staff in this category should be the major
producers of research outputs, and the main supervisors of doctoral students. Graph
10 shows how the totals of permanent academic staff with doctoral degrees changed
over the period 1996 - 2010.
Graph 10: total permanent academic staff: 1996-2010
18000
16000
14000
13449
14184
14673
15423
15809
15936
5146
5403
16684
13098
12000
10000
8000
6000
4000
4647
4658
4561
4572
4485
5957
2000
0
1996
1998
2000
2002
2004
Doctorate as highest qualification
2006
2008
2010
Total permanent
14
15 Graph 11 divides public HE institutions into the 3 categories used for national
planning purposes, and sub-divides the 11 universities into a group of 6 which
produces 60% of the HE system’s total high level knowledge products and the
remaining 5 [The differentiation analysis in Part III splits these groups into three
clusters based on a wider range of indicators]:
High productive universities:
UCT, UKZN, Pretoria, Rhodes, Stellenbosch, Wits;
Oher universities:
Fort Hare, Free State, Limpopo, North West, UWC
Comprehensive universities:
Johannesburg, NMMU, Unisa, Venda, WSU,
Zululand
Universities of technology:
Cape Peninsula, Central, Durban, Mangosuthu,
Tshwane, Vaal, DUT
16 Graph 12 relates research publications to total academic staff and to academic
staff with doctorates. It shows what the ratios were between total research
publications (see Graph 1) and (a) total permanent academics and (b) academics
with doctorates. The years selected are 2004 (1st year of the new funding
15
framework), and (b) 2010 (latest available data).
Graph 11:
% of academic staff with doctorates by
institutional category
60%
50%
48%
45%
41%
40%
30%
36%
36%
29%
29%
44%
40%
37%
28%
35%
28%
44%
40%
38%
29%
28%
20%
15%
13%
10%
5%
7%
8%
10%
0%
2000
2002
2004
High productive universities
Comprehenives
2006
2008
2010
Other universities
Universities of technology
16
Additional graph A: distribution of permanent academic staff by rank: 2010
70%
59%
60%
59%
58%
50%
50%
41%
40%
34%
30%
34%
28%
25%
20%
16%
23%
25%
25%
30%
28%
25%
18%
41%
40%
38%
32%
26%
27%
UWC
WITS
17%
10%
0%
UCT
FORT HARE
UKZN
Professors + assoc professors
NMMU
NWU
Senior lecturers
TUT
Lecturers + junior lecturers
17
Additional graph B: proportions of academic staff with doctorates by rank: 2010
100%
90%
80%
70%
92%
83%
81%
77%
66%
60%
50%
40%
84%
82%
62%
61%
59%
49%
44%
36%
91%
40%
62%
49%
49%
49%
38%
33%
28%
30%
14%
20%
58%
53%
17%
13%
10%
27%
19%
9%
13%
3%
0%
UCT
FORT HARE
Professors + assoc professors
UKZN
NMMU
Senior lecturers
NWU
TUT
Lecturers + junior lecturers
UWC
WITS
Total permanent academics
18
Graph 12:
research publications per permanent academic by
institutional category
0.17
0.44
0.46
All acads
2010
0.93
1.10
1.61
With doct
1.15
1.95
0.07
All acads
0.30
0.33
2004
0.73
0.80
1.06
With doct
1.21
1.83
0.00
0.50
1.00
1.50
2.00
High productive universities
Other universities
Comprehensives
Universities of technology
2.50
19
PART II: GOVERNMENT RESEARCH FUNDING
RESEARCH FUNDING INCENTIVES
Graph 13:
Graph 14:
Graph 15:
Subsidised research outputs of the public higher education system: 2000 - 2010
Government research funding allocations by output category and financial year:
Rands millions
Estimates of Rand values of research output units: Rands thousands
INSTITUTIONAL CAPACITY AND RESEARCH OUTPUTS
Graph 16:
Graph 17:
Government research output funding by institutional category: Rands millions
Total government research output funding per permanent academic: Rands
thousands
INSTITUTIONAL RESPONSES TO DOCTORAL INCENTIVES
Graph 18:
Graph 19:
Doctoral graduate funding by institutional category: Rands millions
Doctoral & publication output funding per permanent academic in 2011/12:
Rands thousands
20
RESEARCH FUNDING INCENTIVES
17 Government’s funding incentives for research outputs are complex because of
the two-year time lag between the completing of an output and the receipt of a
funding allocation, the weightings applied to research outputs, and the way in
which the research output budget is determined:
(a) The output funding for the 2004/5 financial year was based on the outputs for
2002, and the 2004 outputs (the first actually generated under the new framework)
would have received research funding only in the 2006/7.
(b) Doctoral graduates are given a weighting of 3, research publications a weighting
of 1, and research masters graduates (which included in the calculations) also a
weighting of 1.
(c) The annual government budget for research outputs is determined as a % of the
total government budget for public higher education, and is not generated by
actual research outputs. So the starting point for the division of output funds for
2004/5 was a provision of R845 million, for 2006/7 a provision of R1 237 million,
and for 2011/12 a provision of R2 225 million.
21
18
The totals in Graph 13 below and the budget allocations can be used to
calculate (a) what research funding was generated by each research output
category, and (b) a Rand value for each output unit for a given financial year. These
data can be seen in Graph 14 and Graph 15 which follow on the next two slides.
Graph 13: subsidised research outputs of the public higher education
system: 2000 - 2010
12000
10000
9748
8086
8074
8000
6660
6000
5602
5476
4000
3181
2728
3421
3857
3723
1100
1182
4647
1421
2000
969
961
1104
0
2000
2002
Publication units
2004
2006
Research masters graduates
2008
2010
Doctoral graduates
22
Graph 14: government research funding allocations by output category and
financial year: Rands millions
2500
2225
2000
1837
1540
1500
1245
1237
1000
845
500
919
1225
1048
1024
796
474
489
505
192
179
228
202
265
253
652
596
343
310
282
298
379
539
461
414
375
365
0
2004/05
2005/06
Publications units
2006/07
2007/08
2008/09
Research masters grads
2009/10
2010/11
Doctoral graduates
2011/12
TOTAL
23
Graph 15: estimates of Rand values of research output units
(Rands thousands)
450
400
393
350
300
286
251
250
200
191
150
100
50
87
66
91
82
110
102
134
110
0
2005/06
Per publication unit
2007/08
2009/10
Per research masters graduate
2011/12
Per doctoral graduate
24
19
The high Rand value of each doctoral graduate is a consequence of the
weighting of 3. The growth data in Graph 2 suggest however that this financial
incentive has not yet affected doctoral graduate growth, which was 3.5% pa
between 2000 & 2004, and 3.6% pa between 2005 and 2010.
20
There are likely to be a number of reasons why doctoral graduate totals
have not yet responded to the output funding incentives introduced for the first time
in the 2004/5 financial year. One explanation is that doctoral processes in SA have
been characterised by the high levels of inefficiency of the kind highlighted in
Graphs 6 to 9.
21
A further explanation that only a few universities have had the capacity to
benefit from the introduction of government research output incentives. This will be
discussed in the slides and paragraphs which follow.
25
INSTITUTIONAL CAPACITY AND RESEARCH OUTPUTS
22
This subsection uses government research funding allocations as proxies
for an institutional capacity to undertake high level knowledge production activities.
The groups will be the government planning categories used earlier in this
presentation:
High productive universities:
UCT, UKZN, Pretoria, Rhodes, Stellenbosch, Wits;
Oher universities :
Fort Hare, Free State, Limpopo, North West, UWC
Comprehensive universities :
Johannesburg, NMMU, Unisa, Venda, WSU,
Zululand
Universities of technology :
Cape Peninsula, Central, Durban, Mangosuthu,
Tshwane, Vaal, DUT
23 Graph 16 shows that the research outputs of the six high productive
universities generated R1 365 million in government funding in 2011/12. This
amount was 63% of the research output total available in 2011/12.
26
Graph 16: government research output funding by institutional
category: Rands millions
1600
1400
1365 (63%)
1200
1000
955
800
785
600
561
400
200
0
419 (19%)
300 (14%)
288
20
162
160
2005/06
188
203
36
2007/08
High productive universities
Comprehenives
249
69 (3%)
51
2009/10
2011/12
Other universities
Universities of technology
27
24
The different research output funding levels reflected in Graph 16 can be
related directly to the capacity of the academic staff in these groupings. For
example, the group of 6 high productive universities has 38% of the total
academic staff in the HE system, and generates over 60% of research output
funding, and the remaining 5 universities have 20% of academic staff and 19% of
research output grants. Comprehensive universities have a 24% share of
academic staff and a 14% share of research output grants. Universities of
technology have a 17% share of academic staff and a 3% share of research
output funding.
25
Graph 17 shows in a slightly different way how government output funding
can be related to staff capacity. The graph shows that in 2011/12 the high
productive universities generated R290 000 in government research funds per
permanent academic, compared to R130 000 for other universities, R66 000 for
comprehensives, and R25 000 per permanent academic for universities of
technology.
28
Graph 17: total government research output funding per permanent
academic (Rands thousands)
350
300
290
250
215
200
150
192
135
130
100
94
50
58
39
68
46
60
66
13
19
25
0
8
2007/08
2009/10
2011/12
2005/06
High productive universities
Other universities
Comprehenives
Universities of technology
29
INSTITUTIONAL RESPONSES TO DOCTORAL INCENTIVES
26
A pattern similar to that in Graph 17 emerges when the research output
funds generated by doctoral graduates are related to institutional categories. These
amounts appear in Graph 18 below.
Graph 18: doctoral graduate funding by institutional category:
Rands millions
350
322 (60% )
300
250
216
200
183
150
100
50
0
113
77
43
41
2
2005/06
High productive universities
50
53
8
66
2007/08
2009/10
Other universities
9
Comprehensives
117 (22% )
84 (16% )
15 (3% )
2011/12
Universities of technology
30
27
Graph 19 relates doctoral graduate funding to permanent academic
staff, but also compares this doctoral funding to research publication funding per
permanent academic. The research output funds generated by research
masters graduates have not been included.
28
The graph shows that in 2011/12 universities in the high productive
group generated R82 000 in doctoral funding per permanent academic, and
R126 000 in research publications. Similar wide differences can be seen in the
other three institutional categories.
29
These lower amounts generated by doctoral graduates could be related
to both institutional inefficiencies, but also to the incentives employed by
institutions. Some institutions distribute publication output funds to authors, but
few (if any) distribute doctoral graduate funds to supervisors. Academic staff
members are therefore likely to gain more direct personal benefits from research
publications than from doctoral graduates.
31
Graph 19 doctoral & publication output funding per
permanent academic in 2011/12: Rands thousands
140
126
120
100
82
80
61
60
42
40
40
20
5
5
11
0
High productive
universities
Other universities
Doctorates
Comprehensives
Universities of
technology
Publications
32
PART III: UCT & INSTITUTIONAL DIFFERENTIATION
ACADEMIC CORE AS BASIS FOR DIFFERENTIATION
Graph 20:
Clustering universities on high level knowledge inputs and outputs
Table 1:
High level knowledge clusters
Table 2:
Elements defining the high level academic core of universities
Table 3:
High level knowledge inputs and outputs in the academic core
DOCTORAL ENTRANTS OF 2002 AND 2003: SUMMARY OF PROGRESS
Graph 21
UCT: graduates & drop outs from 2002 + 2003 intakes of new doctoral
students
Graph 22
UCT: summary of drop outs from 2002 & 2003 cohorts of new doctoral
students
Graph 23
UCT: summary of graduates from 2002 & 2003 cohorts of new doctoral
students
33
ACADEMIC CORES AS BASIS FOR DIFFERENTIATION
30 CHET has used sets of high level knowledge indicators, which form their
academic cores, to place SA’s 23 universities into three distinct clusters. The
universities which appear in these clusters are listed on the next slide.
Graph 20: clustering universities on high level knowledge inputs and outputs
2
1.5
CLUSTER 1
1
0.5
CLUSTER 2
0
-0.5
CLUSTER 3
-1
-1.5
Masters
enrol %
Doctors
enrol %
Staff PhD %
Ratio PhD
enrol to
Staff
M grad rate
PhD grad
rate
Ratio Phd Publication
grad to staff
output
34
31 The universities in the three high level knowledge clusters are set out in the
table below. The clusters differ from the groupings used earlier in the presentation:
• UKZN and Pretoria are not included in Cluster 1 the top producer category
• Cluster 2 consists of a mix of 6 universities and 4 comprehensives
• Cluster 3 contains 1 university, 1 comprehensive and 6 universities of technology
Table 1: high level knowledge clusters
Cluster 1:
top level
producers
Cape Town
Rhodes
Stellenbosch
Witwatersrand
Cluster 2: medium
level producers
Fort Hare
Free State
Johannesburg
KwaZulu-Natal
Nelson Mandela
North West
Pretoria
Unisa
UWC
Zululand
Cluster 3:
low level producers
Cape Peninsula UoT
Central UoT
Durban UoT
Limpopo
Mangosuthu UoT
Venda
Vaal UoT
Walter Sisulu
35
32 CHET has defined the academic core of a university as (a) the set of inputs
which it requires to support its teaching and research activities, and (b) as the set
of outputs which it produces on the basis of these inputs. Table 2 on the next slide
deals with a specific aspect of the academic core of universities: the inputs and
outputs required for high level knowledge production.
33 The indicators in Table 2 differ slightly from those used for the cluster analysis
in Graph 20. Table 2 combines masters and doctoral enrolments into a single
indicator, and adds an additional input indicator. This is research funding available
per permanent academic staff member.
34 Table 3 (on the slide following Table 2) sets out the academic core indicator
values for all 23 universities.
35 Table 3 shows that UCT falls below one of the academic core input targets and
one of the output targets. These relate to doctoral enrolment and graduation rates.
36
37
38
DOCTORAL ENTRANTS OF 2002 AND 2003: SUMMARY OF PROGRESS
36 Graphs 21 – 23 use UCT’s doctoral entering cohorts of 2002 and 2003 to
illustrate some aspects of its doctoral throughput rates.
Graph 21
UCT: graduates & drop outs from 2002 + 2003
intakes of new doctoral students
350
287
300
250
200
180
(63%)
191
125
(65%)
150
107
84
100
66
50
5
1
4 (20%)
7
5
2 (71%)
35
49
(58%)
0
SET
BUS
Total new intake
ED
Drop outs
HUM
TOTAL
Graduates
39
Graph 22
UCT: summary of drop outs from 2002 & 2003 cohorts of
new doctoral students
45%
42%
40%
37%
35%
35%
30%
26%
25%
20%
14%
15%
10%
9%
8%
6% 5% 6%
7%
5%
10%
9%
2%
0%
8%
4%
1%
0%
Dropped year 1 Dropped year 2 Dropped years Dropped year 5
3-4
SET
HUM
Dropped after
year 5
Total drop outs
AVERAGE
40
Graph 23 UCT: summary of graduates from 2002 & 2003 cohorts of new
doctoral students
70%
65%
63%
58%
60%
50%
40%
26%
30%
23%
18%
20%
15%15%15%
10%11%10%
14%13%14%
10%
1% 1% 1%
0%
Graduated years Graduated year Graduated year Graduated years Graduated after Total graduates
1-3
4
5
6-7
year 7
SET
HUM
AVERAGE
41
CONCLUDING NOTES TO PARTS I & II & III
42
37 The higher education system’s high level knowledge outputs are doctoral
graduates, research masters graduates, and research publication. These
increased over the period 2000 to 2010, but differences appeared after the
introduction of a new government funding framework in 2004. Doctoral
graduates grew at a lower rate than the other two categories between 2005 and
2010, probably because of the incentives in the government funding framework
and the limited academic staff capacity in many universities.
38 SA’s problems with academic staff capacity are particularly evident in the low
number and % of academics with doctoral degrees. In 2010 only 6 000 of the
total of 16 700 permanent academics had doctoral degrees, and this group must
have carried primary responsibility for the 12 000 doctoral students enrolled in
the higher education system in 2010. The ratio of 2 doctoral enrolments per
permanent academic with a doctoral degree is probably an appropriate norm.
43
39 Institutional reactions to government funding of research outputs tends to favour
publications rather than doctoral graduates. There do not appear to be strong
incentives for academic staff to focus on doctoral supervision rather than the
production of research articles.
40 The academic core analyses have shown that UCT is in the top cluster of 4 high
level knowledge producers. UCT fell short of the core targets only in respect of
the doctoral student to academic staff ratio, and the graduation rate of doctoral
students. The other three universities were well below target in these areas:
•
Stellenbosch: masters graduate throughputs, doctoral graduate throughputs;
•
Wits:
masters graduate throughputs, doctoral graduate throughputs; ratio of
doctoral graduates to academic staff ;
•
Rhodes:
ratio of doctoral enrolments to academic staff, doctoral graduate
throughputs; ratio of doctoral graduates to academic staff.
44