Budget projection errors:
causes, consequences and
correctives
Melbourne Economic Forum
John Daley and Danielle Wood, Grattan Institute
30 May 2017
Budget projection errors
Problems
• Reality has not matched repeated projections of return to budget balance
• Projections of growth in nominal GDP, wages, and revenues were too high
• Asymmetrical projections – risks not evenly balanced
Causes
• Assume the world hasn’t changed
• Assume short economic cycle: generates over-optimism in bad times
• Amplify optimism by assuming outperformance to close output gap
Consequences
• Governments have forecast-led denial about need for budget repair
• Budget repair relies on bracket creep rather than measures
• Realpolitic suggests we need a pessimism-generator
Correctives
• More judgement and common sense
• Key modelling issues: what is “trend” (average what period?); how fast do
we return to trend; is spare capacity absorbed?
• What are the institutional barriers to change?
2
Projected surpluses have not happened
Actual and projected Commonwealth underlying cash balance, % of GDP
1
Projection made in 2010 2011 2012 2013
2017
2014
0
2015 2016
-1
-2
-3
Actual
-4
-5
2009
2011
2013
2015
2017
2019
2021
Financial year ending
Source: Grattan analysis of Commonwealth Budget Papers 2010-11 to 2017-18
3
Forecasts consistently over-pessimistic
before GFC; over-optimistic after
Actual and forecast nominal GDP growth
10
8
Forecast (1 year ahead)
Actual
6
4
2
0
-2
-4
GFC
1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015
Financial year ending
Source: Grattan analysis of Commonwealth Budget Papers 2007-08 to 2017-18; Chart 3 Statement 7, Budget Paper No.1, 2016-17 Budget
4
2017 budget doubles down: increases
forecast wage growth as the reality goes lower
Actual and projected growth in wages, per cent
4.5
Projection made in
4.0
2011
2010
2017
2012
3.5
2016
2013
2015
2014
3.0
2.5
2.0
Actual
If wages
instead grow
at 2%, tax
collections
will be $12.6
billion lower
in 2021
1.5
2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
Financial year ending
Notes: Total hourly rates of pay excluding bonuses, private and public wages
Source: Grattan analysis of Commonwealth Budget Papers 2010-11 to 2017-18
5
Fiscal projections have missed particularly
for company tax and capital gains tax
Actual and projected tax receipts, $ billions
Personal income tax
100
260
2015 2016
2014
2013
80
2012
2017
2011
220
180
60
Actual
140
100
2010 2012 2014 2016 2018 2020
25
20
15
5
2014
2016
2018
2017
Actual
2014
2011 2012
2013
10
5
0
2012
2016
Capital gains tax
2013
2014 2015
2016
2017
10
201120122013
2015
2014
40
2010 2012 2014 2016 2018 2020
Superannuation taxes
15
Company tax
2020
Source: Grattan analysis of Commonwealth Budget Papers 2010-11 to 2017-18
Actual
2015
2017
2016
Includes capital
gains tax from
individuals,
companies and
super funds
0
2010 2012 2014 2016 2018 2020
Financial year ending
6
Budget projection errors
Problems
• Reality has not matched repeated projections of return to budget balance
• Projections of growth in nominal GDP, wages, and revenues too high
• Asymmetrical projections – risks not evenly balanced
Causes
• Assume the world hasn’t changed
• Assume short economic cycle: generates over-optimism in bad times
• Amplify optimism by assuming outperformance to close output gap
Consequences
• Governments have forecast-led denial about need for budget repair
• Budget repair relies on bracket creep rather than measures
• Realpolitic suggests we need a pessimism-generator
Correctives
• More judgement and common sense
• Key modelling issues: what is “trend” (average what period?); how fast do
we return to trend; is spare capacity absorbed?
• What are the institutional barriers to change?
7
The disappearing and supposedly
reappearing Phillips curve
Growth in employee compensation (%)
20
15
10
Pre-1990s
1990-2009
Budget
forecasts
5
PostGFC
0
3
4
5
6
7
8
9
10
11
12
Unemployment rate (%)
Notes: Employee compensation is total remuneration of employees including wages, ad hoc bonuses, termination payments and in-kind benefits
Source: Based on NAB analysis of ABS Labour Force data
8
Budget projection errors
Problems
• Reality has not matched repeated projections of return to budget balance
• Projections of growth in nominal GDP, wages, and revenues too high
• Asymmetrical projections – risks not evenly balanced
Causes
• Assume the world hasn’t changed
• Assume short economic cycle: generates over-optimism in bad times
• Amplify optimism by assuming outperformance to close output gap
Consequences
• Governments have forecast-led denial about need for budget repair
• Budget repair relies on bracket creep rather than measures
• Realpolitic suggests we need a pessimism-generator
Correctives
• More judgement and common sense
• Key modelling issues: what is “trend” (average what period?); how fast do
we return to trend; is spare capacity absorbed?
• What are the institutional barriers to change?
9
Forecast-led denial is a common affliction
This budget
keeps us on
a sustainable
path to bring
the budget
back to
balance.
We are on track for
surplus in 2012-13, on
time, as promised — and
this provides the solid
foundations for the
targeted investments we
announce tonight.
I can report
tonight that
despite the
headwinds, our
timetable back to
a Budget surplus
is unchanged from
last year.
Sources: Treasurer's budget speech 2011-12; 2015-16; 2016-17
10
Most of the planned structural repair
depends on fiscal drag
Budget balance
2020-21 budget outcome compared to 2016-17 outcome, $2017 billions
20
10
Denotes improvement
in budget balance
7
0
3
-10
-20
5
38
37
-30
-40
7
7
-50
Budget Initial deficit
deficit 2016- growth at
17
nominal GDP
Fiscal
drag
Other
revenue
Spending
Other
2016-17
Budget
restraint (Future fund, Budget deficit 2020non-fin asset measures
21
Variance from GDP growth
11
impacts)
Budgets have included some structural
repair measures, but not many
Contribution of budget measures to fiscal balance at end of four years
$bn budget measures
20
Net impact
Revenue measures
15
Spending measures
10
5
0
-5
-10
-15
-20
-25
2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Year of budget announcement
Source: Grattan analysis of historical Budget Papers, MYEFO and PEFOs
12
Budget projection errors
Problems
• Reality has not matched repeated projections of return to budget balance
• Projections of growth in nominal GDP, wages, and revenues too high
• Asymmetrical projections – risks not evenly balanced
Causes
• Assume the world hasn’t changed
• Assume short economic cycle: generates over-optimism in bad times
• Amplify optimism by assuming outperformance to close output gap
Consequences
• Governments have forecast-led denial about need for budget repair
• Budget repair relies on bracket creep rather than measures
• Realpolitic suggests we need a pessimism-generator
Correctives
• More judgement and common sense
• Key modelling issues: what is “trend” (average what period?); how fast do
we return to trend; is spare capacity absorbed?
• What are the institutional barriers to change?
13
Reviews have not led to agile reform
Errors
Forecast model
Review of Treasury • Nominal GDP and taxation revenue
projections: little evidence of bias
macroeconomic
over two decades
and revenue
forecasting (2012) • BUT forecast errors correlated with
economic cycle
• Recommended
new economy-wide
model to be
embedded into
forecasting
• Too rigorous adherence to models –
Review of Treasury
not enough judgement
macroeconomic
• Risk that short-term forecasts are
forecasting
anchored by long run projections
capabilities (2015)
• Endorsed
recommendation for
economy-wide
forecasting model
• Forecasts should be adjusted based
on judgement – views of sectoral
experts, surveys, business liaison,
views of senior staff
• Models should use historical
evidence of adjustment process
rather than mechanical assumption
• Endorsed
recommendation for
economy-wide
forecasting model
• Macro-econometric
model should span
forecast and
projection period
Review of
economic
modelling at
Treasury (2017)
14
Other nations use more judgement and are
less optimistic
Real GDP growth
Wages growth
4%
4%
2021 (projection)
2017
3%
3%
2%
2%
1%
1%
0%
0%
Aust Canada NZ
UK
US
(OBR) (CBO)
Aust
NZ
UK
US
(OBR) (CBO)
Source: Grattan analysis of latest economic projections by CBO (US), OBR (UK), NZ Treasury and Canadian Department of Finance
15