Accepted Manuscript Further investigations of the donor-flight response Rohan Sweeney, Duncan Mortimer, David W. Johnston PII:

S0277-9536(14)00279-2

DOI:

10.1016/j.socscimed.2014.04.041

Reference:

SSM 9443

To appear in:

Social Science & Medicine

Received Date: 7 April 2014 Accepted Date: 30 April 2014

Please cite this article as: Sweeney, R., Mortimer, D., Johnston, D.W., Further investigations of the donor-flight response, Social Science & Medicine (2014), doi: 10.1016/j.socscimed.2014.04.041. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Further investigations of the donor-flight response. Introduction In Sweeney et al. (2014), we presented quantitative evidence that SWAp implementation has been associated with a 29.4% reduction in DAH receipts amongst a poorest subgroup of implementing

RI PT

countries. This evidence was derived using rigorous econometric methods and best available data. In this paper, we discuss the contribution such methods can make in assessing impacts of health SWAps and further consider issues of heterogeneity and measurement error. Additional regression

SC

results are presented that demonstrate the key findings in Sweeney et al. (2014), are robust to changes in model assumptions, including those suggested by Paul et al. Finally, we discuss

M AN U

interpretation of our findings, test an alternative explanation of the cause of donor-flight and propose future research to strengthen the evidence base on impacts of the health SWAp.

TE D

The value of panel regression techniques

McNee (2012) warns of a range of measurement issues that “reduce the power” of many findings presented in previous SWAp reviews and evaluation papers. This is a literature dominated by case-

EP

study evaluations. Case studies make an important contribution in certain circumstances, such as assessment of perceived government engagement with and stewardship of individual SWAps.

AC C

However, they have significant limitations when we try to quantify the effects of the SWAp approach, including its impact on DAH flows or health outcomes. At a minimum, quantification of SWAp effects requires a comparable, contemporaneous control group to limit confounding. And the potential confounding effects are noteworthy. The period of SWAp take-up has coincided with unprecedented increases in DAH as well as exposure to the Aid Effectiveness Agenda promoting harmonisation and coordination of all countries’ aid programmes regardless of SWAp status (OECD, 2010; Ravishankar et al., 2009). Randomised control trials (RCTs) are not possible. In the absence of RCTs, the econometric methods utilised in Sweeney et al. (2014) represent the best available means of

ACCEPTED MANUSCRIPT comparing SWAp against a contemporaneous control and therefore makes a valuable contribution to understanding SWAp effects. While the econometric analysis offers certain advantages over case studies, there is a risk of over interpreting findings from both approaches. Our study estimated the average treatment effect

RI PT

across a sample of DAH recipients, not the effect of SWAp implementation in each country analysed. The donor-flight effect identified in the poorest subgroup of implementers will be higher in some countries and lower, or even absent, in others. It should also be emphasised that this average

SC

treatment effect does not necessarily imply average absolute reductions in overall levels of DAH received. Rather, comparison against a contemporaneous control group suggests that, on average,

they not implemented a SWAp.

TE D

Heterogeneity

M AN U

SWAp-implementing countries appear to have received less DAH than they might have received had

SWAp is an evolving process. Further, SWAp is not a homogenous treatment. On this we are in agreement with Paul et al. Indeed, we explicitly recognised that “SWAps have been operationalised

EP

differently in different countries” and conducted additional analyses to understand its potential impact (Sweeney et al., 2014). Specifically, we undertook subgroup analyses in early implementing

AC C

and poorest subgroups and explored temporal variation in the SWAp treatment effect (using dummy variables to identify SWAps at different stages of implementation). The decision to code SWAp as a binary variable was driven by the data (or lack thereof) on health SWAps. Setting aside data limitations, a country’s SWAp status might instead be captured by a continuous index of ‘SWAp completeness’, measuring the stage of development and level of government/donor participation in an implementing country at a given point in time. Alternatively, several indices might be required to capture continuous variation in implementation of SWAp

ACCEPTED MANUSCRIPT components. The availability of continuous indices would permit exploration of threshold effects and the contribution of the component parts of SWAp to the average SWAp effect. However, the lack of such indices by no means makes the average effects estimated in our study uninformative. Countries with implemented health SWAps have commenced a formalised process of aid

RI PT

coordination and harmonisation, distinguishing them from countries that have not chosen to

implement. The modelled SWAp variable can be likened to a dummy variable that captures intention to treat rather than treatment per protocol. The ‘dose’ of treatment varies due to operationalization

SC

differences or levels of participation by government and donors. Moreover, results derived from our approach are perfectly consistent with the interpretation of SWAps as an evolving process if we

M AN U

consider the SWAp dummy variable as an indicator of the beginning of this process. The analysis of evolving SWAp effects (SWAp 1-2yrs prior; SWAp 1-2 yrs post; SWAp 2+yrs post) describe

Measurement error

TE D

development of the SWAp through discrete periods of implementation and maturation.

Establishing a dataset of SWAp implementation was challenging as “there is no formal international

EP

record-keeping on SWAps” (Peters et al., 2013)(p.3). Further, opinions differ on when a SWAp becomes a SWAp (Walford, 2003). The Web-Table [insert web-link here] attached sets out evidence

AC C

identified to support a country’s assigned positive SWAp status and commencement year. As noted in the Web-Table [insert web-link here] and argued by Paul et al., the SWAp status and timing of implementation for some countries is uncertain. Where measurement error with respect to SWAp status is present in a limited number of potentially influential observations, robustness analyses presented in Sweeney et al. (2014) demonstrated our main result was not due to miscoding of individual SWAps.

ACCEPTED MANUSCRIPT Additional tests of sensitivity to uncertain SWAp status and commencement assumptions are presented below. First, we test sensitivity of the average donor-flight treatment effect to plausible variation in commencement year. FIGURE 1 presents the set of average estimated donor-flight effects after systematically commencing each country’s SWAp, both one year earlier and one year

RI PT

later for both the main sample and the poorest subgroup of countries. These are compared to the estimated effects presented in the original analysis represented by vertical lines at -16.6% (for the main sample) and -29.4% (for the poorest subgroup). The final row in Figure 1 reports average

SC

treatment effects after moving SWAp commencement times for all SWAp implementers one year forward or backward in a single estimation.

Poorest subgroup

Main sample

-16.6% estimated effect

EP

TE D

- 29.4% estimated effect

AC C

Bangladesh Burkina Faso Burundi Ethiopia Ghana Lesotho Madagascar Malawi Mali Mozambique Nepal Nicaragua Niger Papua New Guinea Rwanda Samoa Solomon Is. Sudan Tanzania Uganda Vietnam Zambia ALL COUNTRIES

M AN U

FIGURE 1.Tests of sensitivity to SWAp commencement assumptions

-30% -28%

-20% -18% -16% -14%

-10%

% reduction in DAH

assumes SWAp commenced 1 year earlier assumes SWAp commenced 1 year later

In the main sample, results appear most sensitive to an earlier commencement of SWAp in Papua New Guinea and Samoa, and in the poorest subgroup of SWAp implementers the result is most sensitive to an earlier commencement of the Mozambique and Niger SWAps. Overall, the estimated

ACCEPTED MANUSCRIPT effects are quite stable in these sensitivity analyses and even the largest magnitude changes are qualitatively consistent with results from our original analyses. In addition, all re-estimations in the poorest subgroup sensitivity analysis (including the all-countries re-estimation), remain statistically significant at the 10% level. The evidence that SWAp is associated with a reduction of DAH in poorest

RI PT

implementers compared to the counterfactual therefore remains compelling. Of course, more systematic errors in coding SWAps would bias our estimate of the average

treatment effect. However, we are confident that this type of systematic miscoding has not occurred.

SC

Almost all identified countries with SWAp were labelled as such in the International Health

Partnerships Plus Country Planning Cycle Database (IHP+, 2011)(IHP+ 2011)(IHP+, 2011). The

M AN U

attached Web-Table [insert web-link here] provides further supportive evidence of positive SWAp status and assigned commencement years. That said, to investigate the potential impact of specific concerns raised by Paul et al., we re-estimated the impact of SWAp implementation on DAH when Burkina Faso and Madagascar are excluded from the analysis. Paul et al. further specified concern

TE D

about Mauritania’s assigned commencement and Congo DRCs non-SWAp status. Mauritania was excluded from all samples in the propensity-to-SWAp selection process (Sweeney et al., 2014), thus did not impact on results. The IHP+ CPC database stated the Democratic Republic of Congo (DRC)

EP

had no SWAp at the time of the search (IHP+, 2011) and this remained the case at the time of writing this paper (March 2014). Further, Rothmann et al. (2011) reported DRC did not have the building

AC C

blocks in place for SWAp during field visits that took place in 2009-10, suggesting progress identified by Zinnen (2011) happened post the timeframe of our analysis. Nonetheless, DRC has also been removed from the control group in a further test of robustness. The associated re-estimated SWAp effects are presented in TABLE 1, and show reductions in the magnitude of the effect size in both samples. Yet the impact of SWAp on DAH delivered to the poorest subgroup remains significant at the 10% level. Again, there remains compelling evidence that SWAp implementation has been associated with a significant reduction in DAH compared to the counterfactual.

ACCEPTED MANUSCRIPT TABLE 1. Estimated impact of SWAp on log-levels of DAH – Sensitivity to Exclusion of SWAp Observations a

Coefficient (p value)

% change in DAH

Main sample -0.181 (p= 0.21)

-16.56%

Burkina Faso & Madagascar excluded

-0.163 (p=0.29)

-15.03%

Burkina Faso, Madagascar & DRC excluded

-0.129 (p=0.40)

Poorest sample -0.348 (p=0.03)

Burkina Faso & Madagascar excluded

-0.341 (p=0.06)

Burkina Faso, Madagascar & DRC excluded

-0.292 (p=0.099)

-12.08%

-29.39%

-28.87%

-25.35%

Halvorsen & Palmquist (1980) adjustment to coefficient for dummy variable estimator in semi-logarithmic model.

SC

a

Original estimate (Sweeney et al., 2014)

RI PT

Original estimate (Sweeney et al., 2014)

M AN U

Alternative hypotheses, interpretation of findings and next steps

In Sweeney et al. (2014), we hypothesised the main drivers of reduced DAH after SWAp implementation were misalignment of donor and recipient government priorities and lack of confidence in recipient government systems. Paul et al. offer an alternative explanation. Specifically,

TE D

that increases in general ODA (not tagged as DAH) compensate for reductions in DAH and that poorer countries are systematically more likely to receive compensating increases in general ODA. Certainly, such an effect could partly or wholly account for the more pronounced donor-flight effect

EP

in the poorest subsample. We test this explanation in supplementary regressions using all ODA

AC C

(including DAH) as the dependent variable in place of DAH. We found that SWAp implementation was associated with a non-significant 18% reduction in total ODA (p=0.15) in the poorest subgroup. While this result is (broadly) consistent with a compensatory increase in general ODA (not tagged as DAH), further research would be required to evaluate the net effect on delivery of health services and health outcomes. In particular, the flow through of any increase in general ODA to the health sector may be somewhat less than the reduction in DAH. Paul et al. note that, following SWAp implementation in Mali, external funding for health has effectively increased and been more efficiently delivered via increased general budget support. However, this may not be the experience

ACCEPTED MANUSCRIPT of all SWAp implementing countries. Fernandes Antunes et al. (2013) recently found that increased general budget support has not, on average, increased government spending on health from both domestic and external sources. In their commentary, Paul et al suggest clearer interpretation of findings might be obtained if DAH

RI PT

tagged as general sector support was used to reflect donor participation in SWAp. However, initial hopes that SWAp would lead all donors to pool DAH and deliver as general sector support have not been fulfilled. Health SWAps now regularly encompass project specific funding, so long as funded

SC

activities fall under a country’s strategic health plan, agreed upon within a SWAp (Chansa et al., 2008; Hill, 2002; Walford, 2003). Thus disentangling project and programme aid would not help separate

M AN U

out SWAp DAH for an isolated analysis. Further, we believe investigating a potential donor-flight effect is made more meaningful by including all DAH from all donors engaged with a given country. It is true that implemented health SWAps have generally achieved only mixed levels of donor participation (Hill, 2002; Jefferys & Walford, 2003; Walford, 2003). However, this does not mean

TE D

that non-participating donors are indifferent to, or unaffected by, SWAp implementation (Hill, 2002; Peters et al., 2013). As Hill (2002) writes, “making participation [in SWAp] optional creates a line of demarcation around which donors can strategically deploy themselves, providing new opportunities

EP

for leverage both within the group, and outside it” (p.1731).

AC C

Finally, Paul et al. warn of the dangers of a far more extreme position than we take in our paper. Specifically, that our work somewhat questions whether poorest countries “should bother trying to better coordinate aid”. This was neither a finding of our analysis nor a point raised in our discussion. The motivation for our analysis and discussion of our findings explicitly recognise that there may be a trade-off between the effectiveness of SWAp and the quantum of aid such that a smaller pot of more effective aid may produce greater health gains than a larger pot of less effective aid. While we acknowledge that more aid may not always be better, we would also argue that less aid may

ACCEPTED MANUSCRIPT sometimes be a bad thing and that there is (as yet) little evidence for a causal relationship between SWAp and improvements in population health (McNee, 2012).

We believe these analyses provide compelling evidence that SWAp implementation has been

RI PT

associated with a reduction in DAH in poorest implementing countries. Of course, uncertainties exist. Additional, comprehensive sensitivity analyses testing the importance of these uncertainties have been presented. The identified donor-flight effect remains. We have made considerable progress in

SC

controlling for confounding utilising appropriate panel data regressions, yet our methods identify only average effects across a set of heterogeneous health SWAps. As recognised above, a more

M AN U

sensitive measure of SWAp status will further enhance our ability to accurately assess impacts of health SWAp implementation. A positive next step would draw on development practitioners and health policy experts engaged in health SWAps to develop an international database on health SWAps in implementing countries - documenting the depth and breadth of SWAp components and

EP

References

TE D

where possible their timing. And we’d like to help.

Chansa, C., Sundewall, J., McIntyre, D., Tomson, G., & Forsberg, B.C. (2008). Exploring SWAps

AC C

contribution to the efficient allocation and use of resources in the health sector in Zambia. Health Policy and Planning, 23, 244-251.

Fernandes Antunes, A., Xu, K., James, C.D., Saksena, P., Van de Maele, N., Carrin, G., et al. (2013). GENERAL BUDGET SUPPORT: HAS IT BENEFITED THE HEALTH SECTOR? Health Economics, 22, 1440-1451. Hill, P.S. (2002). The rhetoric of sector-wide approaches for health development. Social Science & Medicine, 54, 1725-1737.

ACCEPTED MANUSCRIPT IHP+. (2011). Country planning cycle database: a WHO resource. International Health Partnerships. Retrieved 14/03/2012, from http://www.internationalhealthpartnership.net/en/tools/country-planning-database/. Jefferys, E., & Walford, V. (2003). Mapping of sector wide approaches; a report for the Swedish

RI PT

International Development Cooperation Agency. London: Institute for Health Sector Development.

McNee, A. (2012). Rethinking health sector wide approaches through the lens of aid effectiveness. In

SC

Development Policy Centre - Crawford School of Public Policy (Ed.), Discussion Paper: Australian National University.

Directorate (Ed.): OECD.

M AN U

OECD. (2010). Paris Declaration and Accra Agenda for Action. In OECD Development Co-operation

Peters, D.H., Paina, L., & Schleimann, F. (2013). Sector-wide approaches (SWAps) in health: what have we learned? Health Policy and Planning, 28, 884-890.

TE D

Ravishankar, N., Gubbins, P., Cooley, R.J., Leach-Kemon, K., Michaud, C.M., Jamison, D.T., et al. (2009). Financing of global health: tracking development assistance for health from 1990 to 2007. The Lancet, 373, 2113-2124.

EP

Rothmann, I., Canavan, A., Cassimon, D., Coolen, A., & Verbeke, K. (2011). Moving towards a sectorwide approach (SWAp) for health in fragile states: lessons learned on the state of readiness

AC C

in Timor Leste, Sierra Leone and Democratic Republic of Congo. KIT Working Paper Series. http://www.kit.nl/net/KIT_Publicaties_output/ShowFile2.aspx?e=1749. Last accessed

27/03/2014.

Sweeney, R., Mortimer, D., & Johnston, D.W. (2014). Do Sector Wide Approaches for health aid delivery lead to ‘donor-flight’? A comparison of 46 low-income countries. Social Science & Medicine, 105, 38-46.

ACCEPTED MANUSCRIPT Vaillancourt, D. (2009). Do health sector-wide approaches achieve results? In Independent Evaluation Group (Ed.), Independent Evaluation Group Working Paper Series Document number 2009/4. Washington DC: The World Bank. Walford, V. (2003). Defining and evaluating SWAps: a paper for the Inter-Agency Group on SWAps

AC C

EP

TE D

M AN U

SC

RI PT

and Development Cooperation. Technical Paper No. 29. HLSP Institute. London.

ACCEPTED MANUSCRIPT

Supportive evidence

• • • Benin

2010

• •

Burkina Faso

2001

• •

• Burundi

2008



Decision

1997 has been assigned as it represents the earlier estimate of SWAp implementation.

SC

1997

M AN U

Bangladesh

Described as with SWAp in IHP+ CPC Database (IHP+, 2011). Jefferys and Walford (2003) state Bangladesh’s health SWAp began in 1997. Negin and Hort (2010) state the Bangladesh SWAp commenced in 1998. Draft statement of intent between Benin Ministry of Health and development partners (IHP+, 2011), indicating likely pre-SWAp status. But further investigations warranted. Smith et al. (2005) explicitly stated in report on Benin health system no SWAp was present in 2005. In 2008 Paul et al. (2008) described that in Benin the idea of a health SWAp was “only nascent” with little interest from the government. World Bank Integrated Safeguards Data Sheet for a Benin Health System Performance Project set to commence in 2010, stated a Project Development Objective was to prepare foundations for SWAp implementation in Benin (World Bank, 2009). Described as with SWAp in IHP+ CPC Database (IHP+, 2011). Jefferys and Walford (2003) outline the Burkina Faso health SWAp was officially adopted in 2001 in multi-country study, though describes implementation proper beginning in 2002. Not declared a SWAp implementer in IHP+ CPC Database (IHP+, 2011). Other documentation suggests SWAp developments underway. Memorandum of Understanding reportedly signed by the Government and health partners in February 2008, for implementing IHP “within the framework of the sector-wide

TE D

• •

EP

Year implemented

AC C

SWAp country

RI PT

Web Table 1: Supportive evidence of Country health SWAp status and year of commencement.

World Bank (2009) suggested concerted effort towards SWAp planned for 2010. 2010 has thus been assigned for a Benin SWAp in the database, however sensitivity tests of inclusion should be conducted where relevant due to uncertain SWAp status. Note, countries with an assigned commencement of 2010, which immediately followed the period of analysed data in Sweeney et al. (2014), were excluded from propensity score matching. This was to avoid the possibility of including a SWAp in the control group (due to potential uncertainty around the exact commencement date). 2001 has been assigned as it represents earlier estimate of SWAp implementation.

Plausible that early stage of SWAp implementation commenced in 2008, however sensitivity tests of inclusion should be conducted where relevant due to uncertain SWAp status.

1

ACCEPTED MANUSCRIPT

• • •

1997



Ghana

1997

• • • •

2005

Lesotho

2005



• •

M AN U

Ethiopia

TE D

• •

EP



Kyrgyzstan

SWiM’s limited fund pooling remains directly comparable to SWAp where partial fund pooling is now common (Jefferys & Walford, 2003; Negin & Hort, 2010). Thus Cambodia has been included as with SWAp in the dataset. Note, Cambodia had insufficient data (including incomplete baseline observation) for inclusion in analyses in Sweeney et al. (2014).

SC

2000

The IHP+ CPC database reported Cambodia had “support to develop a SWAp”(IHP+, 2011). Cambodia has Sector Wide Management (SWiM) in place with plans to move to a full SWAp (Lane, 2007). SWiM is reportedly very comparable to SWAp, with only partial fund pooling, and a focus on donor harmonisation and alignment (Negin & Hort, 2010; OECD, 2006). It is effectively considered a SWAp in some research (Hutton & Tanner, 2004; Negin & Hort, 2010). Negin and Hort (2010) state SWiM in Cambodia commenced in 2000. Described as with SWAp in IHP+ CPC Database (IHP+, 2011). No specific documentation stating SWAp commencement time was found. The country’s first Health Sector Development Programme (HSDP 1) commenced in 1997 and, along with education, reportedly had most complete sector programme with donor support (Worku, 2002; World Bank, 2001). Described as with SWAp in IHP+ CPC Database (IHP+, 2011). Jefferys and Walford (2003) state Ghana’s health SWAp began in 1997. Described as with SWAp in IHP+ CPC Database (IHP+, 2011). The World Bank International Development Association reportedly led a consortium of donors who agreed to adopt a health SWAp in 2004 (World Bank IDA, 2008). In a presentation by Cudré-Mauroux and Muratalieva (year unknown) from the Swiss Cooperation Office, it is implied the formalisation of the Kyrgyz Republic’s SWAp occurred around 2005. Described as with SWAp in IHP+ CPC Database (IHP+, 2011). The WHO’s Lesotho Country Cooperation Strategy 2008-2013

AC C

Cambodia

RI PT

approach” and that technical support was needed to assist the design and introduction of a health SWAp between March 2008March 2009 (IHP+, 2008).

Based upon cited documents, a 1997 commencement year has been assigned. Sensitivity tests of commencement should be conducted where relevant due to uncertainty of SWAp commencement.

Based upon cited documents, a 1997 commencement has been assigned. Based upon cited documents, a 2005 commencement year has been assigned. Note, Cambodia had insufficient data (including incomplete baseline observation) for inclusion in analyses in Sweeney et al. (2014).

Based upon cited documents, a 2005 commencement year has been assigned. It likely represents an early stage of

2

ACCEPTED MANUSCRIPT

2007

• • Malawi

2004

• •

Mali

1998

• •

• • Mauritania

2000

RI PT

SWAp implementation.

SC

Given cited World Bank project was to facilitate formal implementation of SWAp, it has been assumed UNICEF status indicated only a plan for SWAp was present in 2005. 2007 has been assigned as commencement date.

M AN U

• Madagascar

TE D

• •

EP



AC C



states that a health SWAp was already in place, in its early phases (WHO, 2009b). A report on UNICEF’s engagement in SWAps indicates that in 2005 Lesotho had a SWAp in the health sector (UNICEF, 2006). A review of Irish Aid to the Lesotho health sector stated in mid2005 a consultancy was engaged to “look into the implementation of a health SWAp”, though went on to indicate the SWAp was in its early stages (Marsden & O'Connell, 2007). Described as with SWAp in IHP+ CPC Database (IHP+, 2011). A report on UNICEF’s engagement in SWAps indicates that in 2005 Lesotho had a health SWAp “planned or under development” (UNICEF, 2006). World Bank Report on Madagascar health SWAp indicated the purpose of upcoming (2007) World Bank project was to facilitate the formal implementation of the health SWAp, stating there was already “general agreement among the Government and a large number of development partners on a sector wide approach to the development and support of the health sector” (World Bank, 2007). Described as with SWAp in IHP+ CPC Database (IHP+, 2011). Bowie and Mwase (2011) state that 2002/2003 was the financial year before the health SWAp started. Report by the Human Development Resource Centre and the UK Department for International Development (DFID) reported that after 2004, the SWAp had been evolving (Pearson, 2010). Walford (2007) describes the Malawi SWAp as in its early stages of implementation in 2004. Described as with SWAp in IHP+ CPC Database (IHP+, 2011). In essay on lessons from SWAp experiences of six African countries, Paul et al. (2008) state that the Mali SWAp commenced in 1998. Described as with SWAp in IHP+ CPC Database (IHP+, 2011) . A report from the World Bank Independent Evaluation Group (2005) states a World Bank led Health Sector Investment Program (HSIP) was approved in the 1998 financial year and implemented

It is not clear from Bowie and Mwase (2011) when in the financial year 2003/2004 the SWAp commenced. Given Walford described the SWAp as in its early stages of implementation in 2004, this commencement year has been assigned.

Based upon cited documents, a 1998 commencement has been assigned.

It is not certain the early implementation of the HSIP was through a SWAp and no further informative documentation were found. Originally, a commencement year of 2000 was assigned, based upon the conservative assumption that Mauritania was not one of the earlier African SWAps

3

ACCEPTED MANUSCRIPT

Mongolia

2009

• • Mozambique

1997 •

Nepal

2004

• •

TE D

• •

Described as with SWAp in IHP+ CPC Database (IHP+, 2011). Ulikpan et al. (2012) describe the stages of Mongolia’s health SWAp evolution. During the period 2005-2009 they described early discussions, predominantly donor interest and SWAp components were only “addressed on paper, not in practice” (p.62). The SWAp reportedly started to evolve when Government ownership increased in 2009. Described as with SWAp in IHP+ CPC Database (IHP+, 2011). Jefferys and Walford (2003) states the Mozambique health SWAp commenced in 1997. Though alternative commencements of 2000 (Martinez, 2006) and 2001 (Walford, 2007) were found. Described as with SWAp in IHP+ CPC Database (IHP+, 2011). The IHP+ Taking Stock Report: Nepal states the health SWAp implementation commenced in 2004 (Nepal Ministry of Health and Population & IHP+, 2008).

EP

2010

AC C

Moldova

The IHP+ CPC database reported Moldova had “support to develop a SWAp” (IHP+, 2011). The WHO/Europe stated that they, an EC delegation and Moldova were expected to sign a SWAp agreement in 2010 (WHO, 2014), though no further evidence was found to confirm if this took place.

M AN U



RI PT



referred to as a group byVaillancourt (2009). However, the establishment of the HSIP in 1998, may suggest a slightly earlier commencement. Thus sensitivity tests of commencement should be conducted where relevant due to uncertainty of SWAp commencement. Note, a low score in the propensity to implement SWAp matching process in Sweeney et al. (2014) resulted in exclusion from analyses. It has been assumed early developments for SWAp have taken place based upon expected signing of SWAp agreement in 2010. Sensitivity tests of inclusion should be conducted where relevant due to uncertain SWAp status. Note, countries with an assigned commencement of 2010, which immediately followed the period of analysed data in Sweeney et al. (2014), were excluded from propensity score matching. This was to avoid the possibility of including a SWAp in the control group (due to potential uncertainty around the exact commencement date). Further, Moldova had insufficient data (including incomplete baseline observation) for inclusion. Based upon cited points, a SWAp commencement year of 2009 was assigned in (Sweeney et al., 2014), though some SWAp components may have been put in place during the period 2005-2009. Thus upon reflection an earlier commencement is plausible. Note, a low propensity to implement SWAp resulted in exclusion from Sweeney et al. (2014) analyses. 2001 has been assigned as it represents the earlier estimate of SWAp implementation.

SC



through a SWAp. This program had support from several donors and NGOs, but did not include pooling of funds (Vaillancourt, 2009; World Bank Independent Evaluation Group, 2005). A footnote in Vaillancourt (2009) indicated that a number of African health SWAps commenced between 1997 and ‘the new millennium”, Mauritania was one of the countries referred to (p.3).

Based upon cited documents, a SWAp commencement year of 2004 was assigned.

4

ACCEPTED MANUSCRIPT

• •

Niger

2006



• • • Papua New Guinea

Rwanda

2008

2007



• •

RI PT

SC



M AN U

2005

TE D

Nicaragua

Based upon cited documents, a SWAp commencement year of 2005 was assigned.

OECD report indicates plans were underway; though no clear indication of early implementation. 2006 was first cited evidence that SWAp was in place, thus 2006 was assigned. However concerns that Niger didn’t have a “real SWAp” implies sensitivity tests of inclusion should be conducted where relevant due to uncertain SWAp status.

In the absence of clearer evidence, 2008 has been assigned. Sensitivity tests of inclusion should be conducted due to uncertain SWAp status/commencement.

EP

• •

Vaillancourt (2009) provides supportive (though non-conclusive) evidence stating the Nepal World Bank Health Sector Program commenced in 2004. Described as with SWAp in IHP+ CPC Database (IHP+, 2011). A World Bank project outline and the OECD Survey on Harmonisation and Alignment of Donor Practice indicate preliminary developments for the SWAp were underway, but it was nascent when the survey was conducted in 2004 (OECD, 2006; World Bank, 2004). UNICEF’s engagement in SWAps indicates that in 2005 Nicaragua had a health SWAp in place (UNICEF, 2006). Described as with SWAp in IHP+ CPC Database (IHP+, 2011). OECD Survey on Harmonisation and Alignment of Donor Practice indicate plans for a sector programme were underway when the survey was conducted in 2004 (OECD, 2006). The Word Bank Booster Program for Malaria Control in Africa describes a World Bank contribution to the Niger SWAp that took place in 2006, indicating it was in place by then (World Bank, 2013). It should be noted that Paul et al. (2008) described it as “not a real SWAp”. The IHP+ CPC database reports that there is “ongoing work to establish SWAp” in PNG (IHP+, 2011). PNG established a pooled account for health funding called the Health Sector Improvement Program Trust Account established in 2003 (Bauze et al., 2009; Sweeney & Mulou, 2012) While dubbed a SWAp, it has been argued by Foster et al. (2009) that until 2008 it was not really a SWAp and could only be considered to be working towards a SWAp. It is not clear if required changes have been made but the cited report implied these next steps were in progress. Described as with SWAp in IHP+ CPC Database (IHP+, 2011). A health SWAp memorandum of understanding was established in 2007 (IHP+, 2011; Rwanda Ministry of Finance and Economic Planning, year unclear)

AC C



Based upon cited documents, a SWAp commencement year of 2007 was assigned.

5

ACCEPTED MANUSCRIPT

• Solomon Islands

Sudan

2009



2006

• • •

• • Tanzania

1997 • • •

Timor Leste

2009 •

RI PT

1997

• •

Whilst IHP+ suggests a pre-SWAp phase, Negin’s more detailed case study suggests early stage of SWAp commenced in 2007. Thus a commencement year of 2007 was assigned. Sensitivity tests of inclusion should be conducted due to uncertain SWAp status/commencement. 1997 has been assigned as it represents the earlier estimate of SWAp implementation.

SC



M AN U

Senegal

2007

TE D

Samoa

EP



Presentations given by the Swiss Agency for Development and Cooperation in May and July, 2007 provide further support that the SWAp was established and in early stages in 2007/2008 (Swiss Agency for Development & Cooperation, 2007a, b). The IHP+ CPC database reports that Samoa has “support to develop a SWAp” (IHP+, 2011). Negin (2010) states the Samoan health SWAp “emerged during 2007” (p.5) and pooled funding under the SWAp was cited in reports dated 2008. Described as with SWAp in IHP+ CPC Database (IHP+, 2011). Jefferys and Walford (2003) states the Senegal health SWAp commenced in 1997, though Watt (2005) indicates it may not have been until later that some donors considered it a SWAp. Not declared a SWAp implementer in IHP+ CPC Database (IHP+, 2011). Negin (2010) describes unsuccessful attempts to introduce SWAp between 2007-2008. Going on to say the SWAp took hold with Government ownership and significant donor participation in 2009. Described as with SWAp in IHP+ CPC Database (IHP+, 2011). No clear evidence regarding commencement time was found. The Sudan Multi-Donor Trust Fund which included funds for the health sector and public health expenditure tracking was established in 2005, with disbursements commencing in 2006 (WHO, 2009a). Described as with SWAp in IHP+ CPC Database (IHP+, 2011). Jefferys and Walford (2003) states the Senegal health SWAp commenced in 1997. Paul et al. (2008) state a 1998 commencement.

AC C



Described as with SWAp in IHP+ CPC Database (IHP+, 2011). Timor Leste’s Health Sector Strategic Plan 2008-2012, sets out the desire to establish a health SWAp (Ministry of Health Timor-Leste, 2007). Rothmann et al. (2011) described Timor Leste as having

Based upon Negin’s detailed case study, a 2009 commencement has been assigned.

The evidence supporting SWAp commencement is weak. Sensitivity tests of inclusion should be conducted due to uncertain SWAp status/commencement.

1997 has been assigned as it represents the earlier estimate of SWAp implementation.

Based upon cited evidence, 2009 likely represents the early stage of SWAp implementation. Sensitivity tests of commencement time should be conducted due to uncertain timing. Note, Timor Leste had insufficient data (including

6

ACCEPTED MANUSCRIPT

Post 2009



Vietnam

2009

• • • •

Zambia

1993

RI PT

The IHP+ database stated Vietnam had an endorsed Statement of Intent in place between the Ministry of Health and development partners in 2009 (IHP+, 2011). This Statement of Intent is the country’s current IHP+ partnership agreement and includes 15 donors. Cox et al. (2011) describe the Statement of Intent as being like a SWAp without complex funding mechanisms. Described as with SWAp in IHP+ CPC Database (IHP+, 2011). Zambia health reforms reportedly started in 1992 ((Jefferys & Walford, 2003), while the SWAp itself reportedly commenced in 1993 (Chansa et al., 2008). Interestingly a formal Memorandum of Understanding was not signed until 1999 (Jefferys & Walford, 2003).

Based upon cited documents, a SWAp commencement year of 1993 was assigned.

AC C



Uzbekistan was originally assigned a commencement of 2008 in (Sweeney et al., 2014) supported by an informal communication with a health systems consultant engaged in Kyrgyzstan. However, the later discovery of the statement by Mirzoev et al. (2010), brings this into question. Note, Uzbekistan had insufficient data (including incomplete baseline observation) for inclusion in analyses in Sweeney et al. (2014). Vietnam has been assigned a commencement of 2009, based upon its signed Statement of Intent. This plausibly represents a pre-SWAp stage, though other reported SWAps also do not have complex funding pools, so may be comparable to those SWAps. Sensitivity tests of inclusion should be conducted due to uncertain SWAp status.

SC

• Uzbekistan

M AN U

• •

TE D

2000

incomplete baseline observation) for inclusion in analyses in Sweeney et al. (2014). Based upon cited documents, a SWAp commencement year of 2000 was assigned.

EP

Uganda

• •

characteristics of an early SWAp including pooled funding during field visit assessment in 2009. Described as with SWAp in IHP+ CPC Database (IHP+, 2011). The Ugandan health SWAp is reported to have commenced in 2000 in multiple sources (Jefferys & Walford, 2003; Walford, 2007). Described as with SWAp in IHP+ CPC Database (IHP+, 2011). No further evidence was found relating to the presence or commencement time of a health SWAp in Uzbekistan. Mirzoev et al. (2010) stated in their 2010 paper that among the former Soviet Union nations in Central Asia, only Kyrgyzstan had formally implemented a SWAp.

Abbreviations: International Health Partnerships Plus (IHP+), Organisation for Economic Co-operation and Development (OECD), World Bank (WB), World Health Organization (WHO)). References:

Bauze, A., Morgan, C., & Kitau, R. (2009). Papua New Guinea: developing an investment case for financing equitable progress towards MDGs 4 and 5 in the Asia Pacific region - phase 1: mapping report. http://www.burnet.edu.au/system/asset/file/273/ICMappingReport-PNG.pdf. Last accessed 27/04/2014. 7

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Bowie, C., & Mwase, T. (2011). Assessing the use of an essential health package in a sector wide approach in Malawi. Health Research Policy and Systems, 9, 4. Chansa, C., Sundewall, J., McIntyre, D., Tomson, G., & Forsberg, B.C. (2008). Exploring SWAps contribution to the efficient allocation and use of resources in the health sector in Zambia. Health Policy and Planning, 23, 244-251. Cox, M., Thi Hanh, T., Hung, T., & Dinh, D. (2011). Paris Declaration/Hanoi core statement phase 2 evaluation: Vietnam country evaluation. http://www.oecd.org/countries/vietnam/47675183.pdf. Last accessed 27/04/2014. Cudré-Mauroux, C., & Muratalieva, E. (year unknown). The health SWAp in the Kyrgyz Republic: . Swiss Cooperation Office & Swiss Tropical and Public Health Institute. Retrieved 04/10/2011, from http://www.swisstph.ch/fileadmin/user_upload/Pdfs/swap/Kyrgystan.pdf. Foster, M., Condon, R., Henderson, S., Janovsky, K., Roche, C., & Slatyer, B. (2009). Working paper 1: Papua New Guinea country report - evaluation of Australian aid to health service delivery in Papua New Guinea, Solomon Islands and Vanuatu. Canberra: Australian Government AusAID Office of Development Effectiveness. Hutton, G., & Tanner, M. (2004). The sector-wide approach: a blessing for public health? Bulletin of the World Health Organization, 82, 893. IHP+. (2008). Taking stock report Burundi: prepared for inter-regional country health sector teams meeting. Report no.6: International Health Partnerships Plus. IHP+. (2011). Country planning cycle database: a WHO resource. International Health Partnerships. Retrieved 14/03/2012, from http://www.internationalhealthpartnership.net/en/tools/country-planning-database/ Jefferys, E., & Walford, V. (2003). Mapping of sector wide approaches; a report for the Swedish International Development Cooperation Agency. London: Institute for Health Sector Development. Lane, C. (2007). Scaling up for better health in Cambodia: a country case study for the World Health Organisation in follow-up to the High Level Forum on the Health Millennium Development Goals. Document no. WHO/HDS/2007.1. World Health Organization, Kingdom of Cambodia Ministry of Health. Marsden, P., & O'Connell, A. (2007). Review of Irish Aid contribution to health sector reform programme in Lesotho. pp. 1-38): Liverpool Associates in Tropical Health. Martinez, J. (2006). Implementing a sector wide approach in health: the case of Mozambique. HLSP Institute technical approach paper pp. 1-8). London: HLSP. Ministry of Health Timor-Leste. (2007). Health sector strategic plan 2008 - 2012. Timor-Leste: Ministry of Health. Mirzoev, T., Green, A., & Newell, J. (2010). Health SWAps and external aid—a case study from Tajikistan. The International Journal of Health Planning and Management, 25, 270-286. Negin, J. (2010). Sector-wide approaches for health: a comparative study of experiences in Samoa and the Solomon Islands. In Health Policy and Health Finance Knowledge Hub (Ed.), Working Paper Series. Vol. 3. Melbourne: The Nossal Institute for Global Health. Negin, J., & Hort, K. (2010). Governance and management arrangements for health sector-wide approaches (SWAps): examples from Africa, Asia and the Pacific. In Health Policy and Health Finance Knowledge Hub (Ed.), Working Paper Series. Vol. 8. Melbourne: The Nossal Institute for Global Health. Nepal Ministry of Health and Population & IHP+. (2008). International health partnership plus: taking stock report: Nepal In International Health Partnership Plus (Ed.), Taking stock report. 8

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

OECD. (2006). Survey on harmonisation and alignment of donor practices: measuring aid harmonisation and alignment in 14 partner countries. OECD DAC Journal pp. 1-191): Organisation for Economic Co-operation and Development. Retrieved 04/10/2011, from http://www.oecdilibrary.org/development/survey-on-harmonisation-and-alignment-of-donor-practices_journal_dev-v6-sup1-en. Paul, C., Paul, E., & Zinnen, V. (2008). Health SWAps: for a sustainable development of healthcare systems in low-income countries. In P.A. Cholewka, & M.M. M. (Eds.), Health capital and sustainable socioeconomic development pp. 191-216. Boca Raton: CRC Press Taylor & Francis Group. Pearson, M. (2010). Impact evaluation of the sector wide approach (SWAp), Malawi: final report. In Human Development Resource Centre (Ed.): Human Development Resource Centre. Document no.273867 / B: 1-75. https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/67670/imp-eval-sect-wde-appr-mw.pdf. Last accessed 27/03/2014. Rothmann, I., Canavan, A., Cassimon, D., Coolen, A., & Verbeke, K. (2011). Moving towards a sector-wide approach (SWAp) for health in fragile states: lessons learned on the state of readiness in Timor Leste, Sierra Leone and Democratic Republic of Congo. KIT Working Paper Series. http://www.kit.nl/net/KIT_Publicaties_output/ShowFile2.aspx?e=1749. Last accessed 27/03/2014. Rwanda Ministry of Finance and Economic Planning. (year unclear). Sector-wide approaches (SWAps) in Rwanda. External Finance Unit, Ministry of Finance and Economic Planning. Retrieved 18/10/2011, 2011, from http://www.devpartners.gov.rw/fileadmin/templates/docs/Studies_and_Reports/Fact%20Sheets/SWAps.pdf. Smith, O., Gbangbade, S., Hounsa, A., & Miller-Franco, L. (2005). Benin: system-wide effects of the Global Fund: interim findings. Bethesda: Partners for Health Reform plus & Abt Associates Inc. http://www.theglobalfund.org/Documents/library/Library_IEPNADD975_Report_en/. Last accessed 27/03/2014. Sweeney, R., Mortimer, D., & Johnston, D.W. (2014). Do Sector Wide Approaches for health aid delivery lead to ‘donor-flight’? A comparison of 46 lowincome countries. Social Science & Medicine, 105, 38-46. Sweeney, R., & Mulou, N. (2012). Fee or free? Trading equity for quality of care for primary health care in Papua New Guinea. International Health, 4, 283288. Swiss Agency for Development & Cooperation. (2007a). Rwanda health SWAp and the role of the Swiss Cooperation in the SWAp development process: SDC health SWAp capitalisation workshop.: Swiss Agency for Development & Cooperation. Retrieved 18/10/2011, 2011, from http://www.swisstph.ch/fileadmin/user_upload/Pdfs/swap/swap470_01.pdf. Swiss Agency for Development & Cooperation. (2007b). Rwanda: health SWAp suggestions & recommendations. Swiss Agency for Development & Cooperation. Ulikpan, A., Hill, P., Malik, A.U., Shakarishvili, G., & Narula, I. (2012). Can a Sector-Wide Approach Underpin and Advance Universal Health Coverage? Asian Social Work and Policy Review, 6, 56-65. UNICEF. (2006). Report on UNICEF engagement in sector-wide approaches. Retrieved 05/10/2011, from http://www.unicef.org/about/execboard/files/2006-8_sectorwide_approahes.pdf. Vaillancourt, D. (2009). Do health sector-wide approaches achieve results? In Independent Evaluation Group (Ed.), Independent Evaluation Group Working Paper Series. Washington DC: The World Bank. 9

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Walford, V. (2007). A review of health sector wide approaches in Africa. In HLSP Institute (Ed.), Technical Paper No. 29. London: HLSP Institute. Watt, P. (2005). Transaction costs in aid: case studies of sector wide approaches in Zambia and Senegal. In UNDP Human Development Report Office (Ed.), Occassional Paper: UNDP Human Development Report Office. WHO. (2009a). Country cooperation strategy for WHO and Sudan 2008-2013. World Health Organization Regional Office for the Eastern Mediterranean. Document no. WHO-EM/ARD/032/E/R/03.09. http://www.who.int/countries/sdn/en/. Last accessed 27/03/2014. WHO. (2009b). WHO country cooperation strategy 2008-2013: Lesotho. World Health Organization Africa Regional Office. Retrieved 04/10/2011, from www.who.int/countryfocus/cooperation_strategy/ccs_lso_en.pdf. WHO. (2014). SWAp in Republic of Moldova. World Health Organization. Retrieved 27/03/2014, from http://www.euro.who.int/en/health-topics/Healthsystems/health-systems-governance/activities/donor-coordination-and-a-sector-wide-approach-to-health-swap/swap-in-republic-of-moldova. Worku, S. (2002). Pooling of technical assistance in the context of sector-wide approaches (SWAps); Ethiopian case study. In European Centre for Development Policy Management (Ed.), ECDPM Discussion Paper. No. 42. Maastricht: ECDPM. World Bank. (2001). Education and health in sub-Saharan Africa: A review of sector-wide approaches. Africa Region Human Development Series, document no. 21905: World Bank. https://openknowledge.worldbank.org/handle/10986/13840. Last accessed 27/03/2014. World Bank. (2004). Republic of Nicaragua health services extension and modernization SWAp Project Information Document (PID) Concept Stage. World Bank. (2007). Madagascar health SWAp project. In W. Bank (Ed.), Integrated Safeguards Data Sheet Concept Stage., Report no: AC2737: World Bank. http://documents.worldbank.org/curated/en/2007/02/7396869/madagascar-health-swap. Last accessed 17/03/2014. World Bank. (2009). Benin: health system performance project: report AC4442. In World Bank (Ed.), integrated safeguards data sheet concept stage. Report no. AC4442: World Bank. World Bank. http://documents.worldbank.org/curated/en/2009/11/11389129/benin-health-system-performance-project. Last accessed 27/03/2014. World Bank. (2013). Niger: institutional strengthening and health sector support project. Washington: The World Bank Group. Retrieved 27/03/2014, from http://go.worldbank.org/S2AY7CFK50. World Bank IDA. (2008). IDA at work: Kyrgyz Republic: government, donors join forces to improve health. World Bank International Development Association. Retrieved October 2011., 2011, from http://web.worldbank.org/WBSITE/EXTERNAL/EXTABOUTUS/IDA/0,,contentMDK:21917849~menuPK:4752068~pagePK:51236175~piPK:437394~th eSitePK:73154,00.html. World Bank Independent Evaluation Group. (2005). Implementation completion report (ICR) review - health sector investment program: Mauritania. In I.E. Group (Ed.): The World Bank Group. Retrieved 2011, from http://lnweb90.worldbank.org/oed/oeddoclib.nsf/DocUNIDViewForJavaSearch/8525682E00686037852570450056BDD9?opendocument.

10

Further investigations of the donor-flight response.

Further investigations of the donor-flight response. - PDF Download Free
586KB Sizes 2 Downloads 3 Views