AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

25
i November 2018 This publication was produced by Nathan Associates Inc. for review by the United States Agency for International Development. AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION CAPACITY AT THE MINISTRY OF FINANCE OF EGYPT Leadership in Public Financial Management II (LPFM II) Selcuk Caner and Tim Robinson

Transcript of AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

Page 1: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

i

November 2018

This publication was produced by Nathan Associates Inc. for review by the United States Agency for International Development.

AN ASSESSMENT OF REVENUE FORECASTING AND

ESTIMATION CAPACITY AT THE MINISTRY OF

FINANCE OF EGYPT

Leadership in Public Financial Management II (LPFM II) Selcuk Caner and Tim Robinson

Page 2: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...
Page 3: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

AN ASSESSMENT OF REVENUE FORECASTING AND

ESTIMATION CAPACITY AT THE MINISTRY OF

FINANCE OF EGYPT

Leadership in Public Financial Management II (LPFM II)

DISCLAIMER

This document is made possible by the support of the American people through the United States Agency for International Development (USAID). Its contents are the sole responsibility of the author or authors and do not necessarily reflect the views of USAID or the United States government.

Page 4: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...
Page 5: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

CONTENTS Acronyms iii

Executive Summary 1

Introduction 5

Revenue Forecasting Practices Worldwide 6

Revenue Forecasting 9

Revenue Estimation 12

Annex 1. Needs by entity 16

Annex II: Data Requests 17

Page 6: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...
Page 7: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

ACRONYMS COR Contracting Officer’s Representative IDIQ Indefinite Delivery Indefinite Quantity Contract LPFM II Leadership in Public Financial Management II PFM Public financial management USAID United States Agency for International Development USAID/E3 United States Agency for International Development Bureau for

Economic Growth, Education and Environment USG United States Government

Page 8: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...
Page 9: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

EXECUTIVE SUMMARY The Ministry of Finance of Egypt (MOFE) is seeking to improve its forecasting and estimation capacity in order to provide more accurate forecasts and assess the revenue impacts of tax policy measures that are intended to mobilize domestic revenues. This assessment considers the techniques the MOFE is currently using, outlines some commonly used strategies in other similar countries, and provides some recommendations to strengthen the current MOFE practices.

Definitions:

Revenue Forecast Impact of macro-economic factors on revenue collection Revenue Estimation Impact of revenue policy changes on revenue collection Economic Impact Analysis Impact of revenue policy changes on the macro-economy Tax Gap Analysis What is the amount we should collect vs what we do collect Tax Expenditure Analysis Estimating the loss of revenue from exemptions

Macro-econometric model

a model estimate the overall economy (including fiscal data) using both regressions and identities to create a consistent framework.

Micro-simulation model a model utilizing tax-payer level data to estimate the impact of changes in policy, or forecast revenue over time from the bottom-up.

Macro-framework model a framework of estimations and identities (similar to the macro-econometric model) creating a consistent estimation framework of the whole economy. (Also known as a “financial programming framework’, or “ratios model”).

This assessment primarily focuses on revenue forecasting and revenue estimation; however, the “economic impact analysis” is also a crucial component to modelling the impact of revenue measures. The “tax gap” and “tax expenditure” analysis can be undertaken through similar models and estimation methods as the revenue estimation.

Overview of roles now

The Government of Egypt has in place tools to conduct macro-level forecasting (a macro-framework model operated by the Macro Fiscal Unit) and a set of mini-models (e.g. for Suez) that feed into this model. Many of the tax lines are estimated through growing historical data by an average historic growth rate, by changes in macro-economic indicators (e.g. inflation or GDP) or by estimating the historic effective tax rate (revenue/revenue base) and forecasting this and the base in to the future. This is in line with standard practice around the world.

Page 10: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 2 A N N E X A

2

The Tax Policy Unit currently conducts analysis of taxes as required, usually focusing around the historic impact of the introduction of new measures and estimation of some new policy impacts (e.g. for increasing tobacco taxes).

Some Revenue Authorities (e.g., Customs, newly established “Economic Analysis Department”) conduct some degree of estimation of the impact of changes; other revenue authorities (tax) produce internal targets for tax inspectorates based on the overall forecast. It is unclear the degree to which forecasts are revised during the year, or the processes for revision (other than updating of the macro-model – which the macro-fiscal unit revises on a regular basis).

Overall, the Government has in place substantial macro-level forecasts but lacks sufficient micro-level forecasts, and does not appear to currently conduct significant monitoring to support re-forecasting during the year or have in place an ex post evaluation process for forecasts.

Recommendations on additional roles in the forecasting and analysis process

- Macro Fiscal unit: macro-level forecasts of revenues.

o Introduce a review process, if not in place, to look back over historic tax forecast errors and identify causes.

o In the new Fiscal Strategy Paper (planned to be introduced) incorporate a write up on changes in forecasts over time (to ensure changes in the forecast are fully justified).

o Review current tax line forecasts to determine if buoyancy or elasticity estimates are suitable, and where new ones can be incorporated.

o Use of more econometric methods including error correction models, which allow a distinction between long and short run effects.

- Tax policy unit: lead on estimation of impact of new policies and micro level forecasts

o Analyze the pattern of collection during the year to support updating the forecast.

o Conduct analysis of the impact of new taxes (using multiple models, including IO models, and micro-simulation models), leading this process in collaboration with revenue authorities, and feeding into the budget forecast estimates.

- Revenue Authorities: supporting micro-level forecasts and estimation of new policies

o Link current work being undertaken (particularly for customs, in assessing the revenue changes from tariff changes in S.M.A.R.T.) to the revenue forecasting process.

o Analyze pattern of collection during the year to determine if tax inspectorates are on track with their targets, and to update their targets.

o Incorporate the use of additional, more customizable models, like TRIST to compliment current work.

The table below outlines the potential expansions of current forecasting that the assessment proposes, by tax type. This paper touches on these models, however they will be explained in more detail during trainings:

Table 1: Tax types, models and the data needed

Page 11: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 3

3

Tax Type Models for forecasting, estimation and analysis

Data Requirement

Personal Income Tax

Micro-simulation model (forecasting and estimation) (A)

Econometric models (forecasting) (B)

Elasticities and Buoyancies (forecasting) (C)

Taxpayer-level micro data (from Tax Authority). (A)

Household income expenditure micro-data (from CAPMAS) (A)

Historic collection data and data on historic macroeconomic variables (B) (C)

Labor force survey summary data (B)

Corporate Tax

Micro-simulation model (forecasting and estimation) (A)

Elasticities and Buoyancies (forecasting) (B)

National Enterprise Survey microdata (from CAPMAS) (A)

Taxpayer-level micro data, or sample (from Tax Authority) (A)

Historic collection data and data on historic macroeconomic variables (B)

VAT Input Output Models (estimation) (A)

Input Output table (from CAPMAS) (A)

Customs Duties

TRIST Model or similar (estimation) (A)

Elasticities and Buoyancies (forecasting) (B)

Estimates of elasticity of response of imports to prices (A)

Historic data on trade (UNCOMTRADE) (A)

Historic collection data and data on historic macroeconomic variables (B)

Fees and Fines AR Models (forecasting) (A) Historic data on fees and fines (A)

Immediate next steps:

1. The government of Egypt is undertaking a major tax reform program as part of the requirements of the IMF program. In order to assess the revenue and economic implications of policy changes, capacity at the tax policy unit (TPU) must be improved with training and additional staff.

2. Firstly, the TPU staff should familiarize themselves with the financial programming methodology used by the Macro-Fiscal Unit. This would require focusing on the revenue side of the financial programming framework. This would enable the TPU to work cooperatively with the Macro-Fiscal Unit and assume some of their responsibilities, and support analysis of the methods used for forecasting the tax heads in the model.

3. Secondly, econometric skills of the TPU staff should be upgraded by introducing them to some more advanced techniques and data adjustments. This would allow them to provide more accurate revenue forecasts.

Page 12: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 4 A N N E X A

4

4. Thirdly, TPU should expand their access to tax and economic data in order to provide more detailed estimates of policy impacts. This would allow the Unit to provide estimates of distributional impacts of policy changes.

Page 13: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 5

5

INTRODUCTION A team from Nathan Associates reviewed the Ministry of Finance of Egypt’s (MOFE) practices in forecasting revenues and estimating revenue impact of revenue proposals during the period September 24 – October 3, 2017. The team met with Vice Minister Amr ElMonayer, Commissioners of Tax Authority Mr. Emad Samy Hussien, Customs Mr. Magdy Abdel Aziz and Real Estate Taxes Dr. Samia Hussein M. Hassan and members of the Macro-Fiscal Unit (MFU) and Tax Policy Analysis Unit (TPU). The team also met with members of the Central Agency of Public Mobilization and Statistics (CAPMAS) to review the availability of economic data in support of revenue forecasting and estimation.

MOFE, the Tax Authority, the Customs Authority and the Real Estate Tax Authority are all engaged in providing revenue forecasts at different levels. MOFE is responsible for providing all budget forecasts; and their forecasts feed in to the pre-budget and budget documentation. The Tax Authority provides forecasts on the expected number of taxpayers and tax revenues for budget projection. These forecasts, and the overall budget forecast, are used as revenue targets for district tax offices. The Customs Authority provides a forecast of the customs duties and other revenues collected by the customs offices for budget projections. Real Estate Tax Authority is not involved in revenue forecasting. But it re-assesses property (rental) values periodically and registers new properties for tax purposes.

MOFE is seeking to improve its forecasting and estimation capacity in order to provide more accurate forecasts and assess the revenue impacts of tax policy measures that are intended to mobilize domestic revenues. The Government’s reform program is expected to increase the tax to GDP ratio from its current level of 12.5 percent by about one percentage point in the next fiscal year. It should be noted that even with new tax measures, the tax to GDP ratio will remain low relative to comparator countries such as Jordan and Turkey.1 However, to inform the potential revenue potential of different tax policy measures, the Government will need tools to more accurately assess their revenue impacts. Assessing revenue and economic impact tax policy measures requires use of more detailed data and new estimation methodologies.

It is important to understand the distinction between revenue forecasting and revenue estimation. Both may require different modeling strategies and data. For example, a time series revenue forecasting model only requires historical data for revenue collections for the tax being modeled. While a time series model may provide accurate forecasts for key macroeconomic and fiscal variables, it may not be possible to estimate the change in revenues resulting from introducing new fiscal policy measures. Revenue and economic impact models are required for measuring the effects of policy changes. Revenue impact models, on the other hand, may require the use of data based on individuals or firms in the case of the personal income tax or enterprise profits tax, respectively. These models also help provide detailed effects on the tax incidence and the distribution of the tax burden.

1 Jordan and Turkey had tax to GDP ratios of 16.7 and 20.5 percent in 2015 respectively.

Page 14: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 6 A N N E X A

6

REVENUE FORECASTING PROCESSES WORLDWIDE Revenue forecasting practices differ around the world, as do the processes by which countries derive their forecast. For example, the UK has an independent body – the Office for Budget Responsibility – which produces a forecast, alongside support from HMRC, the revenue authority. The US has multiple entities, including a congressional body, the Congressional Budget Office Canada produces a forecast based on consensus GDP estimates from banks, reduced by 1%.

There is some variation between countries of different income levels. Within low-income countries (which Egypt does not fall under) practices tend to be concentrated in a single agency (usually the Ministry of Finance)2 . Within OECD countries this responsibility typically falls within the purview of an independent body (e.g. the Netherlands, UK); although still more frequently falls under the Ministry of Finance or similar (e.g. Belgium, France, Ireland, Japan, Italy, New Zealand). It is also somewhat comment in OECD countries, much more so than lower income countries, for a cross agency body to be charged with revenue forecasting3 . The use of independent validation, or external input, is intended to reduce the forecast bias.

The process typically incorporates information from tax collection agencies, macroeconomic units, and other entities. An example process from Finland is set out below (as an example of good practice in separating roles) and Azerbaijan (as an example of a previously disjointed process – improved by the establishment of a working group), as well as the typical process across countries (and split of roles).

Finland

Entities Involved Ministry of Finance: Budget department, Tax department, and Economics Department. Overseen by the Revenue Working Group.

Timing The first tax forecast for the next budget year is put together in the summer before. Following this the forecast is revised during the Economics Department’s forecast rounds and supplementary budget preparations.

2 IMF “Revenue Forecasting—How is it done? Results from a Survey of Low-Income Countries” 3 Revenue Forecasting Practices: Differences across Countries and Consequences for Forecasting Performance

Page 15: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 7

7

Roles Tax department: expert on each line prepares forecast for the individual lines

Macroeconomic department: produces macroforecasts which inform the individual tax lines.

Methods Taxes are estimated based on the performance in a base year, and then estimated changes to the base and any tax policy changes. Behavioural changes are rarely incorporated, unless they can be done with a high degree of accuracy. No revenue is forecast based solely on macro information; the judgement of the individual experts plays a role.

Source and Further Information: “Preparation methods of on-budget revenue estimates in the Ministry of Finance”, Ministry of Finance, Finland

Azerbaijan

Entities Involved Prior to 2003: the Ministry of Finance and Ministry of Tax prepared separate forecasts of revenue; with macroeconomic assumptions from the Ministry of Economic Development. Typically the Minister of Finance would take the higher forecast.

Post 2003: intragovernmental working group was established to improve forecast timing and cohesion.

Timings Prior to 2003: forecasts were finalized very late in the budget planning process, and typically both forecasts were issued initially.

Page 16: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 8 A N N E X A

8

While it is difficult to pull a common approach together across countries, there is a tendency towards an emergent consensus around (for budget revenue forecasting):

Phase Description

Phase 1: Development of Macroeconomic Forecasts

The macroeconomic (macro-fiscal) or similar unit produces a macroeconomic forecast, detailing the likely GDP, CPI etc over the medium term.

Phase 2: estimation of the baseline (i.e. the current year collection)

Estimation or updating of estimates of the current year’s collection. This would typically be done by the tax collection authorities and the tax policy units, and would feed into the macro-fiscal unit’s work.

Phase 3: Development of macro-level forecasts

Macro-level forecasts of higher tax heads (based on simple growth rates, buoyancies and elasticities, or potentially econometric relationships) are developed (typically in Excel based models – even in more advanced countries). Typically, this would be done by either the macro-fiscal unit, or a more specialized tax unit.

Phase 4: Nuancing of macro-level forecasts One-off factors impacting individual tax lines, or changes to the tax forecasting methodology (assessment of previous forecasts for bias etc). This should be done with the tax agencies, tax policy unit in collaboration with the macro fiscal unit (usually through a form of tax forecasting working group).

Phase 5: Micro-level forecasts (and forecasts of lower tax heads)

Estimates of the impact of changes to policies and more detailed estimates of specific, specialist tax lines. This would usually be carried out by the tax policy units, tax collection agencies and potentially any other entities carry out collection of specific taxes (e.g. a mines authority).

Phase 6: Amalgamation of forecast Overseen often by the macro-fiscal unit; forecasts are generated for the budget preparation, and can be adjusted through the process as new tax measures are considered or new economic data emerges.

Page 17: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 9

9

REVENUE FORECASTING CURRENT PRACTICES

Ministry of Finance

The Macro-Fiscal Unit is responsible for forecasting budget revenues and expenditures with a time horizon of one to three years for medium term budgeting purposes. MFU provides forecasts of key revenue and expenditure items. The methodology used is consistent with the macro-fiscal framework used by the International Monetary Fund (IMF) for surveillance and program work4, as well as forecasting systems used in multiple other countries. According to this framework, a baseline scenario is constructed using the trends and growth in key macroeconomic variables. For surveillance purposes, the baseline scenario describes the current state of the economy and its likely path. In countries with IMF-supported programs, the policy scenario, developed jointly with the country authorities, forms the basis for the macroeconomic adjustments. The policy scenario is also used to set specific policy targets that the authorities plans to implement.

According to this framework, GDP and other key macroeconomic variables such as consumption and foreign trade are forecast using single equation regressions where the quantities are expressed as functions of the price level, the interest rate and the money supply. After obtaining the forecast values of GDP, budget revenues are forecast using tax buoyancies, effective tax rates, or trends on fiscal variables.

The Ministry of Finance operates a set of mini-models that feed into the overall macro-framework model (e.g. a model for the Suez canal, utilizing estimates for tonnage etc). Most of the estimations are done through these separate mini-models or through growing historic performance by the average of historic growth rates (or by macroeconomic variables like GDP).

The Ministry of Finance has a Tax Policy Unit which has undertaken some estimation of some tax types (e.g. tobacco tax), to estimate the impact of changes in the rate. This unit sits in a key position within the tax forecast process, and reports directly to the Vice Minister, and needs to be greater utilized (see below sections).

Tax Authority

The economic analysis and statistics units of the Tax Authority provide forecasts of tax revenues by type of tax and region as preliminary budget projections to be compared to similar projections by MFU. Much of the tax authorities’ work involves within-year monitoring of collection, and producing management reports from data. Tax revenues are forecast by growing the historical tax revenue data by average growth rates of tax revenues or estimated GDP growth rates. In addition, the large taxpayer unit (LTU) monitors the top 1,000 taxpayers and their instalment payments in order to determine that the revenues as consistent with budget projections. In addition to forecasting tax revenues, the Tax Authority also provides projections

4 MFU maintains the same tables as the country desk economists at IMF.

Page 18: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 10 A N N E X A

10

for monthly and quarterly cashflows. They also provide internal targets for collectorates, based on the overall tax forecast, and the previous year’s shares of collection by agency.

Customs Authority

The Customs Authority’s Department of Economic Analysis provides forecasts of customs revenues as preliminary estimates for budget projections to be compared with similar projections by the MFU. Growth rates of GDP are used in projecting taxes collected by the Customs Authority. In addition, the Customs Authority also monitors flow of goods entering and leaving the Free Zones. The Customs Authority also conducts studies on the changes in import flows.

Real Estate Tax Authority

The Real Estate Tax Authority collects three types of taxes: 1) real estate tax; 2) land tax; and 3) entertainment tax on various social activities such as theaters amusement parks. The authority also collects a number of fees on behalf of the sub-national governments. In addition to collecting taxes on real property, the Real Estate Tax Authority provides periodic assessment of properties using the rental value approach. The assessment cycle has been reduced to 5 years in 2008. The Authority does not have any capacity in forecasting property tax revenues and it is not considered as one of its activities.

RECOMMENDATIONS

At MOFE, the FPP framework used by MFU is appropriate and should be maintained for forecasting budget revenues. The MFU staff have participated in a number of financial programming and policies (FPP) courses and are well-versed in providing budget revenue forecasts. Therefore, MFU is capable of providing sound revenue forecasts.

Given the nature of the macroeconomic variables used and the recent shocks in the economy, forecasting accuracy may have declined. While there is not much to be done about unanticipated economic shocks, some improvements to the estimation methodology of key macroeconomic variables and tax buoyancy estimates would increase the accuracy of the forecasts.

The MFU can improve forecast accuracy by: 1) revising the estimating models after assessing their historic performance; 2) appropriately adjusting the time series revenue data for discretionary changes in order to estimate tax elasticities accurately; and 3) keeping the past forecast errors and analyzing them for systematic biases. For example, most macroeconomic variables are non-stationary. Therefore, standard regression models would produce biased estimates. As a result, some more appropriate econometric techniques should be considered if not currently used (e.g. Error Correction Models).

For revenue forecasting purposes, these error correction models would be more appropriate where both past values of the dependent variable and the independent variable are used to estimate tax buoyancies. Random policy changes break the relationship between the tax revenues and the macroeconomic variables that are used as proxy for the tax bases. When the tax buoyancies are estimated without this adjustment, the resulting forecast would have larger errors. This can be corrected by adjusting the historical data by the amount of the discretionary change in a proportional manner. Then the adjusted time series data can be used to estimate the

Page 19: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 11

11

true tax elasticity. Use of true elasticities in forecasting models would result in reduced errors. An analysis of errors can eliminate the systematic component of the error terms.

The current institutional arrangement in forecasting budget revenues is consistent with international practice. The existing division of labor in revenue forecasting provides reliable forecasts of budget revenues. Therefore, it is not necessary to undertake any reorganization of the revenue forecasting process. However, it would be useful to share the forecasting methodology used by MFU with the Authorities that also have a stake in providing revenue forecasts. Revenue forecasting capacity at the three revenue collecting authorities can be improved by introducing the same forecasting methodology used by MFU for forecasting underlying economic and fiscal variables. In addition, the forecasting process can be further improved with the use of more advanced estimation techniques recommended for MFU above. These techniques will provide estimates of tax elasticities that will produce improved forecasts.

Page 20: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 12 A N N E X A

12

REVENUE ESTIMATION CURRENT PRACTICE

Ministry of Finance

The revenue forecasting process at MFU provides sufficient detail to produce reliable projections for budget revenues. However, revenue estimation efforts are not yet established. The TPU is assigned with the task of providing such support for policy formulation. However, this unit does not have the resources of the MFU and its capacity to provide tax analysis has to be expanded for effective policy analysis. The TPU has conducted some estimation in the past (e.g. on tobacco tax); however, this work can be expanded.

Tax Authority

There is no explicit revenue estimation at the Tax Authority. However, the Tax Authority undertakes selected studies in order to assess the effects of changing tax legislation. A working group on personal income taxes is collecting taxpayer data to analyze the impact of economic changes on personal income taxes. The Commissioner of Tax Authority is interested in studying the incidence of taxes and its distributional effects using the available taxpayer data. Capacity development is required for the Tax Authority to conduct in-depth analysis of key taxes such as VAT, personal income tax (PIT), and corporate income tax (CIT) including estimation of tax gaps.

Customs Authority

The Customs Authority does not provide any formal revenue impact analysis of changes in economic conditions and legislative changes. However, they have informally investigated the effects of introduction of the value-added tax (VAT) and changes in the mix of imports from consumption goods to manufacturing goods. In addition, using a trade model developed at the World Bank (S.M.A.R.T and a simple CGE model), the Customs Authority experiments with assessing changes in customs revenues and resulting welfare losses. However, the model is not fully calibrated to the case of Egypt. The model has built-in trade elasticities, which may not reflect the response of foreign trade of Egypt to price changes. Nevertheless, there is interest to develop capacity in revenue estimation at the Customs Authority.

Real Estate Tax Authority

There is no revenue estimation effort at the Real Estate Tax Authority. Furthermore, there is no revenue estimation capacity at the Authority and this would most likely not have as urgent a need to undertake this. However, the authority can develop capacity for housing market analysis in the long-run. This analysis would review changes in economic, demographic and housing inventory characteristics of the housing market.

RECOMMENDATIONS

Modeling Requirements

Page 21: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 13

13

Roles of Tax Policy Unit in Forecasting

In broad terms, the functions of tax policy units in the forecasting process are:

1) Monitoring of Revenue Collections;

2) Evaluation of economic, structural and revenue impacts of tax policy options consistent with the key principles of economic efficiency, economic growth, revenue adequacy, revenue stability, simplicity, and low administrative and compliance costs;

3) Tax Expenditure Analysis;

4) Evaluation of the Impact of Non-Tax Economic Policies (e.g. the Investment Law); and

5) Forecasting of Tax Revenues, which includes estimation of tax elasticities, evaluation of changes in economic conditions, and evaluation of the effect of inflation and price changes.

In addition, analysis of the tax incidence and tax expenditures are often undertaken by the TPUs in other countries.

Revenue Estimation Models

Revenue estimation models are fiscal tools and differ from revenue forecasting models both in terms of methodology and data requirements.

There are many different revenue estimation models, each of which has a specific application. One set of revenue estimation models impose the tax structure on the data, which represents the taxpayer population. These are referred to as microsimulation models. They are appropriate to use when estimating the effects of legislative changes on the taxpayer population. These models may be used to simulate policy changes in the taxation of individuals and corporations, respectively. To estimate the revenue impacts of changes to VAT and measure the tax cap, Input-Output modeling would be appropriate. For estimating the revenue and economic effects of various investment incentives, a common methodology is the marginal effective tax rate (METR) methodology. Furthermore, partial or general equilibrium models provide estimates of economy-wide effects of changes in tax policy.

These tools, in addition to providing detailed revenue impact analysis, also can be used in the analysis of tax incidence, estimating the value of tax expenditures provided by the tax system, and tax gap analysis. Results of revenue estimation models can further be used in assessing the collection performance of revenue administration as they provide and an estimate of the potential tax revenue. A summary list if modeling and related data requirements are provided in Table 1 in the Executive Summary.

Microsimulation models

Microsimulation models are based on individual or firm (micro-level) data observations. These models are used as policy analysis tools to address such questions as, “which individuals benefit from tax law changes?”, “which firms pay more or less under new depreciation rules?”, and “which families receive the most benefit from changes in tax credits for children?” These models are based on micro tax return and household survey data. Most microsimulation models use data from tax returns, but other data sources include micro-level (individual-level) surveys of the population including census data and consumer expenditure data.

Page 22: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 14 A N N E X A

14

The benefits of micro-simulation models include the following:

1. models provide detailed estimates of revenues by income group, industry group, type of income, and region;

2. models enable analysts to determine possible areas of evasion behavior by type of income and income class;

3. models provide detailed information on tax bases so changes in tax law are relatively easy to incorporate in the models;

4. models can provide a consistent link among various taxes, especially value added taxes, enterprise profit taxes, and personal income taxes;

5. the design of the models allows policymakers to conduct fiscal analysis; and

6. the analyst can incorporate forecasts from either the univariate or multivariate forecasting models into the micro-simulation model to produce a micro-economic forecast.

Microsimulation models can provide estimates of revenue and incidence of income taxes and amendment options including changes in: 1) rates; 2) inclusion or exclusion of various types of income in the taxable base; 3) exemption amounts and a broad variety of exemption types and, if relevant, phase-out of exemptions; 4) standard deduction amounts and a broad variety of deductions; or 5) credits, such as earned-income and child-care credits, in the case of corporations, the use of investment tax credits and allowances.

Input – Output Models

The newly introduced VAT will likely become the most important revenue source for the budget revenues for Egypt. As a result, analysis of its performance and impacts is crucial and requires development of an input-output model to determine the tax base and the resulting tax revenues by type of industry.

The starting point for constructing the VAT base is the detailed information on domestic consumption, which is obtained by netting the personal and government expenditures abroad from the total expenditures contained in the final demand matrix of the input-output tables.

Then the input output table is adjusted for exempt goods and services, exempt business sales and purchases to determine the taxable VATR base. The input-output model allows for simulating various scenarios such as: 1) change in rates; 2) multiple rates; 3) changes in exemptions; and 5) changes in zero-rating of selected goods and services; as well as estimating the scale of the tax gap by sector.

Marginal effective Tax Rate (METR) methodology

Another important performance criterion of a tax system is its relative efficiency, or to what degree it distorts the allocation of resources in the economy. The efficiency losses from taxation arise because some activities tend to be more heavily taxed than others are, and consequently the allocation of resources gets distorted.

Page 23: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 15

15

METRs measure the tax burden on an investment over the lifetime of the project. The task is to analyze the tax system with an emphasis on compliance and efficiency in administration, and provide a set of options and policy recommendations to improve the efficiency in corporate taxation and in the entire tax system. For this, marginal effective tax rates (METRs) are designed to measure incentives for investment, which are calculations that take into account effects of measurement and timing of income in determining the impact of a tax applied to an additional dollar of capital income. This is particularly relevant for analysis of the tax cost of incentives in of investment and special zones under the new Investment Law.

METR methodology allows for comparison of alternative incentive schemes ranging from tax holidays to accelerated depreciation allowances. METR also provide the information required for a comparison of incentives to invest in different types of assets and different industries. Differences among METRs across assets and industries indicate barriers to efficient allocation of capital among assets and industries since corporations equalize rates of return after corporate taxes.

Data requirements

Microsimulation models

The data for microsimulation models should come from a sample of taxpayers - individuals or corporations - and contain all available information from the tax returns or other reporting documents such as financial statements. Use of taxpayer data allows the model to compute income tax liability according to the laws and regulations. As such, the data is based on personal and corporate income tax returns compiled yearly by the Tax Authority.

In the case of PIT, all taxpayers may not be filing tax returns as their taxes are withheld by their employers. In this case, it would be necessary to augment the taxpayer database with information from household income and expenditure surveys. CIT taxpayer data can be further supplemented with enterprise survey conducted by CAPMAS.

Input Output data

For the analysis of VAT and evaluating the revenue effects, the input-output table prepared by CAPMAS would be required to determine the VAT base. In addition, cells containing sales and purchases in the input-output table have to be adjusted for the exempt goods and services listed in the VAT Law because the input-output table will be at the 2-digit industry level while the list of exemptions refers to goods and services is provided at a more detailed level.

METR

METR methodology does not require an extensive amount of data like the other modeling methodologies. However, depreciation ratios at industry level are required which can be obtained from National Accounts data.

Trade data

Trade data for tax expenditure and revenue estimation purposes can be obtained from the Customs Authority, and should include detailed data by 4-digit HS code on the transactions occurring through the ports (including what tax has been charged – by tax type).

Page 24: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 16 A N N E X A

16

ANNEX 1. NEEDS BY ENTITY

Macro Fiscal Unit

Generates macro level forecast by revenue line:

Historic data grown by macro-indicators

Historic effective tax/base ratios

Mini models for, among other: Suez Canal receipts (utilizes estimates on tonnage etc.)

Tax Policy Unit

Generates forecasts of the impact of new measures and micro level forecasts of revenues (as well as estimates of the tax gap, and tax expenditures):

Occasional estimation of particular tax effects

Macro Fiscal Unit

Additions:

-Assessment of forecasts on a regular basis, looking at biases and methods of improvement (and checking for mis-specification)

- Improved econometric estimates of tax lines (e.g. ECM)

- Elasticities and buoyancy estimations

Tax Policy Unit

Additions:

-Micro-forecasting models for PIT, and CIT based on taxpayer level data (from tax authority databases)

- I-O model for VAT and sales tax estimations based on CAPMAS survey data

- METR analysis

- Estimates of the tax expenditures (using multiple data sources/models)

- Estimates of the tax gap

Customs Authority

Support the tax policy unit in micro-level estimations of revenue and the impact of tariff changes (as well as estimates of tax expenditures and gaps):

Estimation of policy impacts using SMART (as an internal process) Estimation of policy impacts using simple CGE model (as an internal process)

Customs Unit

Additions:

-Inclusion of their work in the budget forecasting process

- Expansion of work on SMART model to include more customizable models (e.g. TRIST)

- Collaboration with TPU on analysis using customs dataset

Tax Authority

Support the tax policy unit in micro-level estimation of revenue and the impact of tax changes (as well as estimates of tax expenditure and gaps):

Regular monitoring of collection

Tax Authority:

Additions:

-Provision of data for micro-forecasting models

- Creation of front-end for databases or standard data requests, for ongoing monitoring of revenue (to support within year re-estimation).

Page 25: AN ASSESSMENT OF REVENUE FORECASTING AND ESTIMATION ...

A- 17

17

ANNEX II: DATA REQUESTS Tax Authority - Taxpayer level micro-data, or samples of

taxpayers (PIT and CIT)

- Collection by major and minor tax type (ideally monthly); including excise by type of item

- VAT collection on domestic VAT by industry code

Customs Authority - Data on transactions at the ports (including value of import, tax paid by tax type and value exempt if any applied) by 4-digit, ideally, HS code from Customs database

- VAT collection on import VAT by industry code

CAPMAS - Household Income Expenditure Survey micro-data

- Labor Force Survey aggregate data

- National Enterprise Survey micro-data or sample of firms (e.g. 1,000 largest firms)