Transcript Thursday Revenue Forecasting

Workshop
Forecasting tax revenues
Forecasting tax revenues
Iris Claus
21 April 2016
Apia, Samoa
Gross domestic product (GDP)
•
… is the income earned from production by
a country’s residents and foreign firms
operating in that country.
Factors
affecting the
economy
•
… is measured and recorded in the
national accounts.
Forecasting
tax revenues
•
… is generally produced and published by
the national statistics office.
•
… can be reported in three ways
Measuring
output
Taxation, GDP
and GNI
Effective tax
rate
approach
Production measure
Expenditure measure
Income measure.
Micro
simulations
Final
comments
Forecasting tax revenues
2 of 35
Production measure
Measuring
output
Taxation, GDP
and GNI
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
•
… is the value added of all resident
institutional units engaged in production.
•
Value added = gross output minus
intermediate consumption
•
… is compiled at a sector or industry level,
e.g. agriculture, food and beverages
manufacturing, hotels and restaurants,
public administration, … .
Forecasting tax revenues
3 of 35
Expenditure measure
Measuring
output
Taxation, GDP
and GNI
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
•
… shows the final use of output.
•
Firms sell output to households and the
government for consumption.
•
Some output is used to expand / replace
the capital stock and / or increase stocks
(inventories).
•
Firms export to other countries.
•
GDP = household Consumption +
Government spending + Investment +
eXports – iMports
•
GDP = C + G + I + X – M
Micro
simulations
Final
comments
Forecasting tax revenues
4 of 35
Income measure
•
Taxation, GDP
and GNI
… is a measure of value added from all
economic activity in a country.
•
Factors
affecting the
economy
(value added = value of output – value of
inputs)
•
… shows the various forms of income
generated by production:
Measuring
output
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Compensation of employees (salaries
and wages)
Operating surplus (firms’ profits)
Consumption of fixed capital
(depreciation)
Indirect taxes less subsidies.
Forecasting tax revenues
5 of 35
Three measures of output
Measuring
output
Taxation, GDP
and GNI
Production
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
GDP
Value added by
industry
Micro
simulations
Expenditure
Income
Consumption
Compensation of
employees
Government
spending
Operating surplus
Investment
Consumption of fixed
capital
Exports
Indirect taxes
less: Imports
less: Subsidies
Final
comments
Forecasting tax revenues
6 of 35
GDP vs gross national income (GNI)
•
GDP is the income earned from production
by a country’s residents and foreign firms
operating in that country.
Factors
affecting the
economy
•
GNI is the income earned by a country’s
residents.
Forecasting
tax revenues
•
It excludes income remitted abroad (e.g.
dividends and interest) and includes similar
income earned by residents from overseas
investments.
Measuring
output
Taxation, GDP
and GNI
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
7 of 35
Taxation, GDP and GNI
Measuring
output
•
A broad based value added tax (VAT)
taxes ‘C’, some ‘I’ (e.g. sale of new owner
occupied dwellings), and some ‘X’ (tourism
spending measured as exports of services).
•
VAT on ‘G’ nets out.
•
In most countries, income taxation seeks to
tax the income generated or the output
produced in a country, i.e. GDP, …
•
… as well as worldwide income of its
residents, i.e. income earned by residents
from overseas investments or parts of GNI.
Taxation, GDP
and GNI
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
8 of 35
Real vs nominal GDP
Measuring
output
Taxation, GDP
and GNI
•
Nominal GDP is a function of quantities and
prices.
•
Real GDP is income or output or value
added in real terms (at constant prices) –
taking into account the effects of price
changes.
•
People and firms make decisions based on
real variables.
•
Tax is collected on nominal income and
expenditure.
•
Tax revenue rises because of increased
economic activity (quantities) or higher
prices.
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
9 of 35
Factors influencing economic activity
Measuring
output
Taxation, GDP
and GNI
•
Temporary business cycle effects
•
Long-run, full capacity, potential output
Factors
affecting the
economy
16
Forecasting
tax revenues
12
Micro
simulations
Final
comments
10
Output
Effective tax
rate
approach
Actual (measured) output
14
8
6
4
Potential output
Cyclical business
cycle effects
2
0
5
10 15 20 25 30 35 40 45
Time
Forecasting tax revenues
10 of 35
Factors influencing potential output
Measuring
output
Taxation, GDP
and GNI
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
•
Productive capacity
•
Capital stock, population growth rate
•
Natural disasters
•
External linkages
•
Financial development, financial stability
•
Property rights, legal system, political
stability
•
Taxation
•
Government expenditure
•
…
Forecasting tax revenues
11 of 35
Factors influencing the business cycle
Measuring
output
Taxation, GDP
and GNI
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
•
External shocks (e.g. global financial crisis,
world food prices, oil prices)
•
Weather
•
Interest rates
•
Exchange rate
•
Fiscal policy
•
Consumer and business confidence
•
…
Final
comments
Forecasting tax revenues
12 of 35
Forecasting tax revenues
Measuring
output
Taxation, GDP
and GNI
•
There are several ways to forecast tax
revenues.
•
The techniques used depend on data
availability as well as the stability of the
relationship between tax revenue and the
variables used to forecast it.
•
We review two approaches:
1.
Effective tax rate approach
2.
Micro simulations
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
13 of 35
Effective tax rate approach
•
Taxation, GDP
and GNI
The tax rate relates the amount of tax
payable to the tax base.
•
Factors
affecting the
economy
The tax base for a given tax is the event or
condition that gives rise to taxation.
•
It is defined in the law and in most cases is
some economic event or condition, e.g.
receipt of wages, sale of goods and
services.
•
The law defines at what rate the event or
condition is taxed, what items may be
deducted in calculating the tax, whether
any exemptions are allowed, the deadline
for paying, the fines that apply to late
payment, etc.
Measuring
output
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
14 of 35
Effective tax rate approach
Measuring
output
Taxation, GDP
and GNI
•
… uses a proxy tax base, i.e. a variable that
is closely related to the actual tax base.
•
… calculates an effective tax rate by
dividing actual tax revenue collected by
the proxy tax base.
•
The effective tax rate would be equal to
the statutory rate …
•
… if we had the perfect proxy tax base, a
single rate, perfect compliance, no
exemptions, no tax free threshold, all
businesses registered for VAT, etc.
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
15 of 35
Proxy tax bases
Measuring
output
Production
Taxation, GDP
and GNI
Factors
affecting the
economy
Value added by
GDP
industry
Forecasting
tax revenues
Effective tax
rate
approach
Expenditure
Income
Consumption
Compensation of
employees
Government
spending
Operating surplus
Investment
Consumption of fixed
capital
Exports
Indirect taxes
less: Imports
less: Subsidies
Tax type
Proxy tax base
Personal income tax
Compensation of employees or total GDP
Micro
simulations
Corporate income tax
Operating surplus or total GDP
Value added tax
Consumption plus exports of services
Final
comments
Excise tax
Consumption plus exports of services
Taxes on international trade
and transactions
Imports
Forecasting tax revenues
16 of 35
Effective tax rates
Measuring
output
Taxation, GDP
and GNI
•
Effective tax rates should be fairly stable
over time.
•
If we have forecasts of GDP (and hence a
proxy tax base), we can use the effective
tax rate to forecast tax revenues
•
… by multiplying the forecast proxy tax
base by the (average) historical effective
tax rate.
•
Some adjustments may be needed to allow
for lags with which taxes are collected (if
taxes are reported on a cash basis).
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
17 of 35
Effective tax rates
Measuring
output
•
Taxation, GDP
and GNI
The effective tax rate may not be stable
and increase or decrease because of:
− Changes in tax policy, e.g. tax rate
changes
Factors
affecting the
economy
− Changes in compliance
Forecasting
tax revenues
− Large tax debts being collected
Effective tax
rate
approach
•
Forecasts will need to be adjusted for these
effects.
Micro
simulations
•
The effective tax rate approach provides a
“top down” approach.
•
Forecast revenue collection is driven by
economic forecasts.
Final
comments
Forecasting tax revenues
18 of 35
Micro simulations
Measuring
output
•
… are based on individual tax return data,
sometimes complemented with survey or
other data.
•
Note: Developing a micro simulation model
is a huge investment. Many tax
administrations and ministries of finance /
treasuries do not have one.
•
But simple micro simulations can still be
undertaken to assess and cost policy
proposals and for revenue forecasting.
•
Micro simulations can produce “bottom
up” tax revenue forecasts to complement
aggregate (effective tax rate) forecasts.
Taxation, GDP
and GNI
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
19 of 35
Micro simulation models
Measuring
output
•
… are computer models that operate at
the level of the individual behavioral entity,
such as a person, family, or firm.
•
… simulate large representative populations
of these low-level entities in order to draw
conclusions that apply to higher levels of
aggregation such as an entire taxpayer
population.
•
… are data intensive, require assumptions
about behavioral responses, linkages in the
economy, etc.
•
… must be sufficienctly detailed to allow
analysis of different scenarios.
•
A decision to develop a micro simulation
model should not be made lightly!
Taxation, GDP
and GNI
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
20 of 35
Micro simulations
•
Simple micro simulations can still be
undertaken to inform policy proposals and
for revenue forecasting.
Factors
affecting the
economy
•
Micro simulations use a representative
sample of taxpayers.
Forecasting
tax revenues
•
Results based on the sample are then
weighted to replicate economy-wide
results.
•
The representative sample is drawn from
the population.
Measuring
output
Taxation, GDP
and GNI
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
21 of 35
Constructing a sample
Measuring
output
•
The first step is identify taxpayer
characteristics (‘dimensions’) to create subgroups (strata) of taxpayers with common
characteristics.
•
E.g. the population of companies may be
stratified (grouped) using the following
dimensions: 25 industry sectors; 10 locations;
3 sizes (large, medium, small); 2 ownership
types (resident, non-resident); 2 types of tax
status (taxable, tax-exempt). These
dimensions would create 3,000
(25x10x3x2x2) sub-groups.
•
Independent sampling of each sub-group
provides a better representation of the
population than random sampling from
non-stratified population data.
Taxation, GDP
and GNI
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
22 of 35
Constructing a sample
Measuring
output
•
The stratified population is used to create
the sample, which has the same number of
sub-groups as the stratified population.
•
Given the significant contribution of large
companies to total corporate income tax
collection, all large corporations should be
in the sample.
•
For each remaining sub-group in the
population, a random sample of firms is
drawn and assigned to the corresponding
sub-group in the sample.
Taxation, GDP
and GNI
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
23 of 35
Validation
Measuring
output
•
The representativeness of the sample needs
to be validated through the comparison of
simulated aggregate outcome with actual
aggregate data.
•
The simulated aggregate outcome is
obtained by weighting the sample.
•
Differences in values between the
population and weighted sample may be
reduced by increasing the sample size and
/ or increasing (or revising) the number of
dimensions used to group taxpayers.
Taxation, GDP
and GNI
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
24 of 35
Weights
Measuring
output
Taxation, GDP
and GNI
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
•
Large firms have a weight of one, as they
are all included in the sample.
•
The other weights are determined by
dividing the total number of firms in a subgroup in the population by the total
number of firms in that sub-group in the
sample.
Agriculture
Manufacturing
Services
Total
Population = 4,010 companies
Large
Medium
Small
2
500
1,700
3
100
100
5
100
1,500
10
700
3,300
Sample = 210 companies
Large
Medium
Small
2
25
85
3
5
5
5
5
75
10
35
165
The sample includes all large companies and 5% of medium and small companies.
Agriculture
Manufacturing
Services
Forecasting tax revenues
Large
1
1
1
Weights
Medium
20
20
20
Small
20
20
20
25 of 35
Potential for micro simulations
Measuring
output
Taxation, GDP
and GNI
•
Micro simulations are a useful tool for
making large datasets more manageable.
•
Micro simulations are more flexible and less
costly than building and maintaining a full
micro simulation model.
•
Just construct a representative sample of
taxpayers for the specific analysis needed.
•
If tax revenue forecasting is the objective,
construct a sample where the simulated
aggregate tax revenue is close to the
actual aggregate tax revenue.
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
26 of 35
Potential for micro simulations
•
Data collection, data management and
data use are investments.
•
They require a vision of what is possible
•
… and a strategy for achieving the vision.
•
They will incur an upfront cost.
•
Effective tax
rate
approach
They will require on-going maintenance.
•
Micro
simulations
The pay-offs from being able to undertake
more rigorous data analysis will be large.
•
Greater use of individual taxpayer data
can enhance designing compliance
strategies, tax policies and forecasting tax
revenues.
Measuring
output
Taxation, GDP
and GNI
Factors
affecting the
economy
Forecasting
tax revenues
Final
comments
Forecasting tax revenues
27 of 35
Potential uses of micro simulations
•
Taxation, GDP
and GNI
What are the characteristics of taxpayers
with outstanding debt?
•
Factors
affecting the
economy
What interventions would be most
effective?
•
How much more tax could we collect?
Forecasting
tax revenues
•
Effective tax
rate
approach
What would be the effect of tightening
rules around entertainment expenditure?
•
Who would be affected by removing
certain tax expenditures?
•
How much more revenue would it raise?
•
What will be the impact of losses carried
forward on future tax collection?
Measuring
output
Micro
simulations
Final
comments
Forecasting tax revenues
28 of 35
Final comments
•
Some taxpayers are not compliant.
•
Some have ‘deep pockets’ – lots of
resources for tax planning.
Factors
affecting the
economy
•
Tax administrations and ministries of finance
/ treasuries have limited resources.
Forecasting
tax revenues
•
We want to use those resources in the best
possible way, complementing what we do
and drawing on our colleagues’ expert
knowledge.
•
The ministry of finance / treasury has
detailed economic forecasts.
•
The tax administration has detailed
taxpayer data and information about
compliance trends.
Measuring
output
Taxation, GDP
and GNI
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
29 of 35
Final comments
Measuring
output
•
If there is information on additional tax
revenue to be collected from compliance
improvement or the settlement of
outstanding debt …
•
… or if revenue losses are expected
because of loop holes in the tax code …
•
… this information needs to be
incorporated into revenue forecasts.
•
Governments need “unbiased” (best
estimates) forecasts for medium-term fiscal
planning.
Taxation, GDP
and GNI
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
30 of 35
Importance of GDP forecasts
Measuring
output
Taxation, GDP
and GNI
•
GDP forecasts are a key input into revenue
forecasts.
•
In Vanuatu, a Technical Committee
prepares estimates of current GDP and
forecasts for the next two years.
•
A Macroeconomic Committee comprising
members from the Ministry of Finance, the
Office of the Prime Minister, the Reserve
Bank of Vanuatu and the Vanuatu National
Statistics Office, approves and publishes
official GDP forecasts.
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
31 of 35
Importance of GDP forecasts
Measuring
output
•
In Tonga, the Ministry of Finance and
National Planning (MFNP) produces
production GDP forecasts, i.e. at the
industry level.
•
The Ministry of Revenue and Customs
(MORC) is setting up a system to forecast
tax revenue collection at the industry level
using the MFNP’s production GDP forecasts
and the effective tax rate approach.
•
Industry forecasts should help identify noncompliance.
•
Non-compliance is detected by collecting
less revenue than expected.
Taxation, GDP
and GNI
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
32 of 35
Importance of GDP forecasts
Measuring
output
Taxation, GDP
and GNI
•
MORC collects and monitors “real time” tax
data.
•
Incorporating actual tax outturns and a
better understanding of tax trends will help
improve tax revenue forecast …
•
… as well as GDP estimates.
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
33 of 35
Importance of GDP forecasts
Measuring
output
•
In Samoa, …
Taxation, GDP
and GNI
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
34 of 35
Discussion points
Measuring
output
•
How do we incorporate information on
compliance, debt collection, revenue losses
into forecasts?
•
What factors are affecting tax collections in
your country? Are the effects temporary or
expected to be long lasting?
•
Is there scope to improve forecasts by
greater sharing of ‘economic and tax
intelligence’? If yes, how?
Taxation, GDP
and GNI
Factors
affecting the
economy
Forecasting
tax revenues
Effective tax
rate
approach
Micro
simulations
Final
comments
Forecasting tax revenues
35 of 35