PUBLIC FINANCE

MTEFs Classification, Developments, Objectives and Impact Assessment Richard Hemming, Duke University
May 1, 2012

MTEF History
Forward estimates and medium-term budgeting in industrial countries MTEFs and PRSPs in 1990s, especially in Africa Part of the World Bank PFM toolkit Donor support World Bank and ODI reviews in early 2000s MTEFs have disappointed because: They ignore initial conditions Weak budget systems Limited institutional capacity There is not enough political support or agency buy in No systematic investigation since, although many diagnoses of why MTEFs do not deliver New World Bank study exploits continued expansion of MTEFs and more years of experience

MTEF Classification
Hierarchy of frameworks defined by highest stage achieved, assuming lower stage(s) is(are) in place Medium-Term Fiscal Framework (MTFF) Resource envelope Agency ceilings Top down, input based Medium-Term Budget Framework (MTBF) National and sector strategies, forward estimates Reconciliation with MTFF, agency/program ceilings Top down and bottom up, primarily input based Medium-Term Performance Framework (MTPF) Systematic use of quantitative performance information Output/outcome based Funding linked to results

Country Classification
Database describing MTEF status of 181countries in every year over the period 1990-2008 This alone is a major contribution of the study Classification is based on key indicators and views of PFM experts Countries are classified 0=no MTEF, 1=MTFF, 2=MTBF, 3=MTPF Identify new MTEFs by stages, transitions between stages, and (only a few) reversals Pilot MTBFs are recorded as MTFFs (but there is some analysis of pilots) Externally imposed multi-year fiscal frameworks (e.g., those underpinning IMF programs) are not recorded at MTFFs

MTEF Adoption

140

120

Number of Countries with MTEFs (line=cumulative, bar=new)

100

80

60

40

20

0 1990 -20 1991-93 1994-96 1997-99 2000-02 2003-05 2006-08

MTEF MTEF

MTFF MTFF

MTBF MTBF

MTPF MTPF

There were 11 MTEFs in 1990, with 1 MTBF (Denmark) and 1 MTPF (Australia. By 2008 there were 132 MTEFs with 71 MTFFs, 42 MTBFs and 19 MTPFs

MTEF Adoption, Country Groups

Number of Countries in Group (percentages of countries in region displayed at top)
16 14 12 10 8 6 16 4 2 0 0 Advanced Economies East Asia & Pacific Europe & Latin America Middle East Central Asia & Caribbean & North Africa South Asia Sub-Saharan Africa 0 0 0 11 16 7 11 25 63 9 39 53 46 37 29 42 43 33 33

MTFF

MTBF

MTPF

MTEF Payoff
Addresses shortcomings of annual budgetingshortsightedness, conservatism (budget rigidities), parochialism (competition for incremental resources) Multi-year planning takes future costs and benefits into account Strategic, forward-looking approach provides a basis for establishing and shifting priorities Collaborative approach to achieving agreed objectives rather than pursuit of narrow self-interest Quality of budgeting improves Contributes to high-level PFM objectives MTFF => + fiscal discipline Testable MTBF => + allocative efficiency hypotheses MTPF => + technical efficiency

Analytical Approaches
Event studies summarize what happened around the time of MTEF implementation Compare averages across countries before and after MTEF implementation date Econometric analysis attempts to explain the impact of MTEFs, controlling for other determinants of fiscal discipline and efficiency Cannot measure everything that may be relevant Nonetheless, some results are very robust But econometric results are empirical regularities, not universal truths Case studies can provide additional insight into the impact of MTEFs, especially insofar as qualitative, nonmeasurable influences are concerned

Measuring Fiscal Performance

Fiscal discipline(1) fiscal balance (MTEF should improve the fiscal balance, with MTFF having the largest effect) Allocative efficiency(2) total expenditure volatility, (3) health expenditure share, (4) health expenditure volatility (MTEF should reduce (2), increase (3) and reduce (4), with MTBF having the largest effect) Technical efficiency(5) cost effectiveness of health expenditure (MTEF should increase (5) with MTPF having the largest effect) Data constraints are significant Data coverage, central vs. general government Expenditure composition, limitations of GFS Expenditure outcomes, health and education

Event Studies
MTEF implementation date normalized at year t Compare three years before and after MTEF implementationcountry and period averages, confidence intervals Data for a maximum of 72 countries, 40 MTFFs, 20 MTBFs and 12 MTPFs Proper interpretationx is higher or y improves after MTEF implementation, not x is higher or y improves due to MTEF implementation (causation), and not even a higher x or improvement in y are related to MTEF implementation (correlation) However, event study findings are supported by econometric results

MTEFs, Fiscal Balance, Expenditure and Revenue

(72 obs)
4 3 2
-0.38

(53 obs)
2 1 0
-1.56

Fiscal Balance

Fiscal Balance

1
0 -1 -2 -3 -4 -5
-3.09

-1 -2 -3 -4 -5

t-5 t-4 t-3 t-2 t-1

t+1 t+2 t+3 t+4 t+5

t-3

t-2

t-1

t+1

t+2

t+3

mtef 95% conf. interval

-6

-3.13

mtef 95% conf. interval

(72 obs) 29 28
Expenditure
25.2

(72 obs)
29 28 27 26 25 24 23 22 21 20
24.8

26 25 24 23 t-3
26.0

Revenue

27

22.9

t-2

t-1

t+1 t+2 t+3

mtef 95% conf. interval

t-3

t-2

t-1

t+1

t+2

t+3

mtef 95% conf. interval

MTFFs and Fiscal Balance (40 obs)

3 2 1 0 -1 -2 -3 -4 -5 mtff 95% conf. interval
-3.06 -0.81

Fiscal Balance

MTPFs and Fiscal Balance (12 obs)

6
2.46

Fiscal Balance
-1.22

2 0 -2 -4 -6
-0.90

MTBFs and Fiscal Balance (20 obs)

4

Fiscal Balance

2 0 -2 -4 -6 -8
-4.46

mtpf 95% conf. interval

mtbf 95% conf. interval

Composition of Fiscal Improvement

Change in fiscal balance = +2.7 percent of GDP Expenditure contribution = -0.8 percent of GDP Revenue contribution = +1.9 percent of GDP What does this say about MTEFs?
Their impact is dominated by other policy changes, such as tax reform Need to focus on expenditure, and especially the ability to implement planned expenditure. They discipline the use of additional revenue, ensuring that it is used for deficit reduction rather than additional spending Other policy changes do matter, but effects are complementary

MTEFs and Efficiency

Total Expenditure Volatility (72 obs)
Total Expenditure Volatility
10 9 8 7 6 5 4 3 2 1 0 t-3
6.1

Health Expenditure Share (72 obs)

15 Health Expenditure Share 14 13 12
11.6

11
10 9 8 7 t-3 t-2 t-1 t t+1 t+2 t+3 mtef 95% conf. interval
11.0

4.7

t-2

t-1

t+1

t+2

t+3

mtef 95% conf. interval

Health Expenditure Volatility (67 obs)

Cost Effectiveness of Health Expenditure 12 Health Expenditure Volatility 10 8 6 4 2 0 t-3 t-2 t-1 t t+1 t+2 t+3
7.2 8.0

Cost Effectiveness of Health Expenditure (41 obs)

94 93 92 91 90 89 88 87 86 85 84 83
90.2

89.9

t-3

t-2

t-1

t+1

t+2

t+3

Econometric Analysis and Results

What can we say about correlation and causality? General model Fiscal performance = f (MTEF status, control variables) Control variables derived from empirical literature on determinants of fiscal performance Serious econometric issues have to be addressed Reverse causalityIs MTEF implementation a response to fiscal performance? Response: Instrumental variables Omitted variablesIs something else going on that affects fiscal performance (and possibly MTEF implementation)? Response: Country and time fixed effects Conditioning variables influence the link between MTEF and fiscal balances Separate regressions for MTFF, MTBF and MTPF adoption (which are 0,1 variables)

Econometric Results
MTEFs have a strong, positive effect on the fiscal balance; the effect gets stronger with move from MTFF to MTBF to MTPF (MTPF has too large an effect) Significant control variablesoil(+), conflict(-), aid(-) Significant conditioning variableOECD(+) for MTPF only MTEFs have a significant positive effect on total expenditure volatility, the health expenditure share and health expenditure volatility; the effect gets stronger with move from MTFF to MTBF to MTPF (except MTPFs have no additional effect on health expenditure volatility) Marginal effect of MTBF over MTFF is surprisingly small Significant control variablesoil, aid, inflation (all + for volatility (because they are volatile?) Only MTPFs have a significant impact on the cost effectiveness of health expenditure Significant control variableinflation(+)

MTEFs and PEFA Scores

Moving Beyond the Data

Empirical work bolsters the case for MTEFs, but they still have to be well designed (coverage, timeframe, disaggregation, status of ceilings and forward estimates, use of margins, institutional responsibilities) What are the broader requirements for effective MTEFs? Commitment to new approach to budgeting Organizational adaptability and technical capacity Appropriate macro-fiscal policies and institutions Sound budget systems and properly sequenced reforms These things have been identified in previous studies Review of Bank experience and case studies can throw new light on these requirements