## 1. Tax-to -GDP Ratio

## Source details

**Canonical URL:** [1. Tax-to -GDP Ratio](https://www.imf.org/-/media/files/publications/cr/2024/english/1kenea2024002.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/cr/2024/english/1kenea2024002.pdf.md)
- [Structured JSON version](/-/media/files/publications/cr/2024/english/1kenea2024002.pdf.json)

---

### Overview and purpose
- Examines tax policy and administrative changes in Eastern African Community (EAC) countries to benchmark Kenya’s experience and draw lessons for future tax reforms.
- Uses granular data from a new IMF database on tax measures announced during 1988–2022.
- High-level conclusion:
  - EAC policymakers frequently changed tax systems and administrations via tax packages that typically combined measures to narrow tax bases (e.g., exemptions, deductions) and to strengthen tax administrative practices (e.g., electronic payments, tax compliance strategy).
- Caveat:
  - Frequency of tax measures is not an indicator of the actual revenue impact of such measures; frequency helps identify reform episodes, duration, and comprehensiveness.

### Kenya: tax-to-GDP trajectory and composition
- Key trajectory points:
  - Peak at 15.5 percent of GDP in 2014.
  - Fell to 13.1 percent of GDP in 2020.
  - Expected to reach 14.4 percent of GDP in 2023.
- Major compositional shifts:
  - Income taxes declined from 8.0 percent of GDP in 2014 to 6.5 percent of GDP in 2020.
  - Tax ratio initially fell during the pandemic (2020–22) in part due to tax breaks, then began rebounding as COVID-19 measures were reversed on January 1, 2022.
  - The 2023 Finance Act is reported to introduce about 1.5 percent of GDP in new tax policy and administrative measures.

### Comparison with other EAC countries
- Kenya is the only EAC country that experienced a protracted fall in its tax-to-GDP ratio over the last decade.
- Except for South Sudan and the Democratic Republic of Congo (oil exporters), other EAC countries have shown either increasing or relatively stable tax-to-GDP ratios over the same period.
- Kenya moved from being the highest among EAC countries in 2012–15 (average values) to being significantly below Rwanda and Burundi in recent years.

### Database, classifications, and coverage
- Database coverage and scope:
  - Covers announced and adopted tax policy and administrative changes in EAC economies over 1988–2022.
  - Source: news clips prepared by tax experts from the International Bureau of Fiscal Documentation.
  - Documents direction of changes (INCREASE/DECREASE) in: RATE, BASE, and ADMIN for seven taxes: personal income tax (PIT), corporate income tax (CIT), value added and sale taxes (VAT), social security contributions and payroll taxes (SSC), excise (EXE), trade taxes (TRADE), and property taxes (PRO).
  - Records announcement and implementation dates (mm/dd/yyyy), category of change (e.g., top rate, exemption, tax compliance), whether part of a broader package, and quantitative information on announced size of most rate changes (generally in percentage points).
- Coverage notes:
  - Database excludes South Sudan due to data limitations and has very limited data coverage for Burundi (2012–17).

### Frequency and composition of tax changes (1988–2022, EAC aggregate)
- Aggregate counts and shares:
  - Total announced changes in EAC countries: 1,845 changes (average about 13 tax policy and administrative changes per year).
  - Composition of changes (share of total identified changes):
    - BASE changes: about 43 percent of total identified changes.
    - ADMIN changes: about 36 percent of total identified changes.
    - RATE changes: remaining portion (about 21.7 percent in table summary).
- Directional patterns:
  - BASE changes: about 60 percent were base-narrowing measures.
  - ADMIN changes: about 80 percent intended to strengthen current practices.
  - RATE changes: announcements of rate hikes were equally frequent as rate reductions.
  - Overall, tax policy measures (RATE and/or BASE) represented about 65 percent of total changes and typically entailed a reduction in taxpayers' liabilities (56 percent of total tax policy changes).

### Kenya-specific frequency and composition (1988–2022)
- Kenya activity:
  - Kenya announced 594 tax policy and administrative changes (average about 18 changes a year).
- Distribution in Kenya (counts and shares as reported):
  - BASE changes: 270 observations (28.2 percent decrease/loosen; 13.0 percent increase/strengthen; total share 41.2 percent).
  - ADMIN changes: 196 observations (5.9 percent decrease/loosen; 23.9 percent increase/strengthen; total share 29.9 percent).
  - RATE changes: 128 observations (9.6 percent decrease/loosen; 9.9 percent increase/strengthen; total share 19.5 percent).
  - Total Kenya: 594 observations (43.8 percent decrease/loosen; 46.8 percent increase/strengthen; combined frequency share 90.5 percent in table).
- Notable Kenya features:
  - Kenya was the only country in the sample where base-narrowing measures were announced more frequently than measures to strengthen administrative practices (31 percent versus 26 percent of total tax measures announced in the country).
  - Kenya was the only country where the frequency of tax policy changes introducing a reduction in taxpayers' liabilities exceeded 60 percent of total tax policy changes (comparison: 52 percent for EAC excluding Kenya).
  - Kenya was most active in announcing changes to CIT and VAT and also most active in announcing changes to PIT, TRADE taxes, and SSC—most of these being BASE changes.

### Cross-country and tax-type patterns
- Aggregate EAC patterns by tax type (share of total tax measures):
  - CIT: 24 percent of total tax measures.
  - VAT: 20 percent of total tax measures.
  - Excise (EXE): 16 percent of total tax measures.
- Nature of changes by tax:
  - CIT and VAT: about half of measures consisted of tax base changes; base-narrowing accounted for the bulk (CIT base-narrowing 60 percent; VAT base-narrowing 68 percent).
  - Excise: 49 percent of excise measures entailed rate changes; excise rate measures were predominantly hikes (75 percent of total excise rate changes).
  - Administrative changes for CIT and VAT accounted for a bit more than one-third of total changes for those taxes; administrative changes for excise were relatively infrequent (13 percent of total excise changes).
  - Measures to strengthen administrative practices were the most common within each tax type (average frequency above 70 percent of respective administrative changes).
- Country differences (high-level):
  - Kenya: frequent base-narrowing for PIT, CIT, VAT, and TRADE; net frequency of base-narrowing exceeded broadening for those taxes.
  - Uganda: most changes to EXCISE and measures to adjust administration of the ensemble of taxes (TAX); EXCISE measures in Uganda appeared to have almost equal shares of RATE and BASE changes.
  - Tanzania: most positive net frequency for EXCISE measures and most negative net frequency for RATE changes for PIT, CIT, VAT, and TRADE.

### Implications, interpretation, and limitations
- Interpretation:
  - Frequent introduction of base-narrowing measures, if sustained over time, can result in an erosion of the tax base and help explain declines in tax-to-GDP ratios (as observed in Kenya since 2014).
  - Policymakers often combined measures to realign rates across taxes, permit some base-narrowing (exemptions/deductions), and improve administrative practices.
- Limitations:
  - Frequency of tax measures does not indicate the magnitude of revenue impact; the database captures counts and directions, not revenue elasticity or fiscal magnitude.
  - Tax outturns are also shaped by macro conditions and behavioral responses, not only by tax decisions.

---

### Most tax changes were announced as part of a broad package of measures

### Prevalence and composition of tax packages (1988–2022)
- Over 1988–2022:
  - Virtually all tax changes (94 percent of total) were announced as part of a broad package of tax measures.
  - Almost 2 out of 3 packages included at least one tax policy measure (tax rate and/or tax base) and one administrative change.
  - Only in 25 percent of the cases were tax rates, tax bases, or administrative measures announced alone.
- Implication:
  - Assessments focusing on an individual tax change will likely be severely biased because policymakers typically changed other tax rates and/or tax bases and/or administrative practices at the same time.

### Co-occurrence of increases and decreases; use of offsets
- Offsetting within packages:
  - Packages with measures lowering tax liabilities were accompanied by at least one measure to increase tax liabilities in 88 percent of total tax packages.
  - Packages with lowering/loosening measures but no accompanying increases/broadening/strengthening represented 12 percent of total packages.
  - EAC countries more often announced packages consisting only of administrative measures to strengthen existing practices without offsetting tax policy measures.
- Implication:
  - Evaluating the revenue impact of a specific tax change requires accounting for simultaneous, potentially offsetting changes in other tax policy and administrative dimensions.

### Time variation and country differences
- Post-Global Financial Crisis: policymakers in EAC countries became more active in changing taxes and/or tax administrations.
- Pandemic (2020–22): most countries intensified interventions to cushion households and businesses and sometimes later repealed previously introduced tax changes (example: Kenya).
- Kenya-specific patterns:
  - Featured the highest number of announcements of tax changes for many years after 2014.
  - Experienced a downward trend in the tax-to-GDP ratio during 2015–22, suggesting investigation of whether announcements tended to reduce taxpayer liabilities.
  - Since 2009 (except 2012 and 2020) showed more measures to strengthen tax administrative practices combined with more measures to reduce tax rates and/or narrow tax bases.
- Uganda-specific patterns:
  - Increased frequency of tax changes significantly from 2017 onwards with tax-to-GDP appearing to improve significantly.
  - Displayed at least six years of co-occurrences of more measures to strengthen tax administrative procedures and raise tax policy (positive net frequencies).
- Cross-country implication:
  - Differing tax performance (e.g., between Kenya and Uganda) may relate to Kenya’s more pronounced tendency to frequently announce tax policy measures that reduce tax rates and narrow tax bases even while strengthening administration.

### Interaction with IMF programs and program-related measures
- EAC countries generally announced more measures to strengthen than relax taxes covered by program conditionality, often accompanied by additional changes in other taxes.
- Kenya examples:
  - Introduced a minimum alternative tax and a digital tax in January 2021 (classified as base-broadening measures and discussed in program negotiations).
  - After approval of the EFF/ECF program on April 2, 2021, authorities:
    - Introduced an exemption from the minimum alternative tax for an air company with a government share above 45 percent and its subsidiaries, and
    - Limited the application of the digital tax to non-residents.
    - These later design changes are captured in the database as CIT base-narrowing measures.
  - VAT base-broadening measures adopted in early 2021 were accompanied in January 2022 by the introduction of several exemptions for medical supplies amid a third wave of COVID-19; these are classified as base-narrowing measures.
- Note:
  - A negative net frequency (more relaxations than strengthenings) does not necessarily imply an overall negative revenue impact; revenue impact requires detailed assessment.

### Key policy implication and recommendation for Kenya
- Kenya needs to strengthen tax collection consistent with authorities’ objectives of sustained increases in tax revenues to meet their development agenda.
- Key milestone: timely adoption of Kenya’s first Medium-Term Revenue Strategy (SR¶17), developed with IMF support.
- Strategy objective: increase revenues by 5 percentage points of GDP by FY2026/27 through measures that broaden the tax base and strengthen tax compliance.

### Data coverage (selected)
- Dataset period referenced throughout analysis: 1988–2022.
- Country data coverage (Appendix I):
  - Kenya 1988-2022
  - Burundi 1989-1995 ; 2004-2017
  - Congo, Democratic Republic of 2001 ; 2009-2022
  - Rwanda 1999-2000 ; 2007-2022
  - Tanzania 1988-2022
  - Uganda 1988-2022

---

### Appendix I: Kenya — Tax policy and administrative changes & PSBS highlights

### Appendix I — selected table highlights (tax change counts and cells)
- The appendix presents detailed counts and cell-level numeric entries for PIT, CIT, VAT, SSC, and EXE as reported in the source (examples preserved exactly as presented):
  - PIT (summary fragments): 99283.535201.864262.5; RATE: 1181.4431.3771.0; Statutory rates: 111.0111.000-; Top rate: 761.2111.0661.0; BASE: 62252.51491.648251.9; ADMIN: 26141.917111.5971.3; etc.
  - CIT (summary fragments): 147314.754163.493293.2; RATE: 23171.4541.318161.1; BASE: 78272.920131.558232.5; ADMIN: 46192.429112.617141.2; etc.
  - VAT (summary fragments): 137284.971262.766252.6; RATE: 28181.612101.216141.1; BASE: 65232.821141.544202.2; ADMIN: 44222.038221.7632.0; etc.
  - SSC (summary fragments): 1061.7961.5111.0; RATE: 321.5321.500-; BASE: 522.5422.0111.0; ADMIN: 221.0221.000-.
  - EXE (summary fragments): 83243.558212.825151.7; RATE: 35172.127151.8871.1; BASE: 30112.71691.814101.4; ADMIN: 18111.61591.7331.0; etc.
- The table distinguishes Decrease / Total changes / Increase while retaining detailed cell-level numeric entries as presented.

### Public Sector Balance Sheet (PSBS): scope and headline metrics (FY2022/23)
- Legal and institutional context:
  - PFMA 2012, amended in 2023, mandates the National Treasury (NT) to be the custodian of an inventory of national government assets and to manage public debt, guarantees and other financial obligations.
  - NT actions: consolidated financial statements for almost all public sector entities; establishment of a Fiscal Risk Committee and a fiscal risk register; ongoing implementation of a treasury single account and accrual-based IPSAS.
- PSBS headline metrics (consolidated, FY2022/23):
  - Total public sector assets: 101 percent of GDP.
  - Public sector estimated financial assets: 14 percent of GDP.
  - Public sector estimated financial liabilities: 131 percent of GDP.
  - Net financial worth: negative 117 percent of GDP.
  - Net worth: minus 30 percent of GDP.
- Additional PSBS details:
  - Public corporations, SAGAs and SCs collectively hold assets worth of 67 percent of GDP as of end-June 2023 (related statement in text).
  - On average, dividends from these entities averaged 0.3 percent of GDP from FY2015/16 to FY2022/23.
  - Dividends remitted by CBK and profitable SCs totaled 0.29 percent of GDP (Table 1 context).
  - In FY2022/23, 242 SAGAs/SCs incurred losses amounting to 0.7 percent of GDP, up from 183 entities with losses equivalent to 0.5 percent of GDP in FY2021/22.
  - As of June 30, 2023, there were 526 SAGAs/SCs, from 500 entities as of end-June 2022, of which 16 have not been included in the consolidated financial statements.
  - In FY2022/23, 18 SAGAs/SCs reported negative equity, totaling 1.5 percent of GDP, up from 1.2 percent of GDP in FY2021/22.
- Compensation and fiscal coverage:
  - National government compensation of employees: declined from 4.3 percent of GDP in FY2019/20 to 3.8 percent of GDP in FY2022/23.
  - SAGAs and SCs compensation of employees: increased by 0.6 percent of GDP over the same period.
  - Entire public sector compensation of employees: remained at about 8.4 percent of GDP.
  - General government compensation of employees estimated at 7 percent of GDP in FY2022/23 or constituting 49 percent of tax revenues.
- Selected PSBS FY2022/23 public sector aggregates (percent of GDP):
  - Entities manage assets worth about 70 percent of GDP.
  - Insurance, pension, and standardized guarantee schemes (liabilities): 33.3 percent of GDP (of which, pension liabilities 33.3 percent of GDP).
  - PPP liabilities: 2.2 percent of GDP.
  - Net financial worth (Public Sector): -117.1 (percent of GDP).
  - Net worth (Public Sector): -30.0 (percent of GDP).
  - Flows (Public Sector): Revenue: 24.0 (percent of GDP); Expenditure: 28.4 (percent of GDP); Net/Lending Borrowing (Profit/Loss): -4.4 (percent of GDP).

### PSBS implications and recommendations
- Broaden fiscal analysis from national government flows to full public sector stocks and flows to better assess fiscal sustainability and risks.
- Use the PSBS to:
  - Improve transparency around assets and liabilities not currently in fiscal reports (natural resources, pension liabilities, PPP contracts, other claims and payables).
  - Strengthen asset management to boost returns, taxes, and dividends from public corporations.
  - Monitor and manage contingent liabilities from counties and state-owned enterprises to reduce unexpected fiscal pressures.
- Continue PFM reforms: consolidate financial statements, implement accrual IPSAS, maintain the Fiscal Risk Committee and fiscal risk register, and complete treasury single account implementation to improve fiscal risk identification and management.
- Medium-term revenue and balance-sheet link:
  - Implementing the medium-term tax strategy (MTRS) is expected to generate additional revenue equivalent to about 5 percent of GDP, enabling strengthening the government balance sheet, less reliance on debt-financing, and increased spending for social policies.
- Arrears and pensions:
  - Clearing government arrears, particularly pending bills, will strengthen PSBS in the medium term.
  - Pension reform (January 2021) converting defined-benefit schemes to defined-contributory is projected to decrease national government pension obligations by about 1 percent of GPD annually for the next 30 to 35 years, but requires proper monitoring and governance.
  - As of end-June 2023, SAGAs and SCs had pension arrears of KSh.46.8 billion, equivalent to 55 percent of the total assets of the new contributory scheme or 12 percent of the combined total assets of the National Social Security Fund and the Public Service Superannuation Fund (new scheme).
- PPPs:
  - Expand and centralize PPP risk management; PPPs may create illusions of fiscal space and carry contingent liabilities. A centralized framework integrating PPP projects into the national public investment and budget framework is essential.

---

### Public corporations, SAGAs/SCs reform and fiscal implications

### Reforms and expected fiscal effects
- Number of SAGAs and SCs rose from 350 in June 2020 to 526 in June 2023, primarily due to the expansion of vocational education and training colleges.
- Authorities are reviewing and rationalizing SAGAs/SCs in line with service delivery mandates.
- A new ownership policy for government owned enterprises will enable an enhanced governance framework for commercial SCs to improve service delivery and profitability.
- Strategic privatization program is anticipated to:
  - generate revenue streams for the government;
  - curtail transfers to non-profitable entities; and
  - alleviate the overall cost of capital borne by public sector assets.
- The reform agenda creates an opportunity to implement distinct governance and monitoring approaches for budget-funded public entities and commercially operated entities.

### Caveats and data limitations
- PSBS is based on published statistics and a few estimates; fixed assets of MDAs have been estimated for FY2022/23 and authorities believe the value of fixed assets is higher than PSBS estimates.
- Cross-holdings of assets and liabilities are likely higher than disclosed; liabilities are estimated from national government debt, pending bills, and SAGAs/SCs liabilities reported by NT.
- PPP contract liabilities use the World Bank database but are presented at half of their stock based on discussions with government officials.
- Pension obligations rely on dated actuarial valuation and assumptions (e.g., non-contributory scheme estimated at 30 percent of GDP; SAGAs/SCs at 3.3 percent of GDP).

### Recommendations and PFM priorities
- Categorize SAGAs and SCs into portfolios (policy and service delivery versus commercial operations) and differentiate them in financial and non-financial terms.
- Conduct an actuarial evaluation for the defined-contribution scheme and monitor the new scheme rigorously.
- Strengthen the PFM system to address accumulation of pending bills, delays in tax refunds, and lack of budget credibility.
- Specific PFM measures: prepare realistic budgets; introduce multi-year commitments for investment projects; execute budgets in accordance with parliamentary appropriations; enhance digitalization of PFM systems; and strengthen procurement processes.
- A credible PFM system is essential to support the debt anchor set at 55 percent of GDP in present value and to prevent accumulation of non-debt liabilities.

---

### Exchange rate passthrough to inflation (historical evidence and policy implications)

### Estimation approaches and data
- Two approaches on quarterly data 1995–2023Q2:
  - Single equation model (local-projections style) using y/y changes: Kenyan inflation, global maize inflation, crude oil prices, changes in nominal exchange rate, and changes in short-term interest rate (three-month T-bill rate).
  - Non-recursive SVAR with two blocks: external variables (global oil prices, U.S. real GDP, U.S. federal funds rate) and domestic block (consumer prices, real GDP, central bank policy rate, three-month T-bill rate, nominal exchange rate); variables in first differences with two lags.
- Data notes:
  - CBK policy rate series from 2003Q3 onward.
  - Exchange rates defined per unit of Kenyan shillings: a positive change is an appreciation of the shilling.

### Main estimation results and headline findings
- Single equation model (y/y changes):
  - On-impact passthrough for a 1 percent change in exchange rate: range between 0.06 and 0.12 (specifications (1), (2), (4), and (5) in Technical Annex Table 2).
  - Cumulative passthrough over a longer period (after lagged inflation): range between 0.22 and 0.42.
  - Passthrough estimates higher for NEER under these specifications.
- Non-recursive SVAR (accumulated impulse responses):
  - Exchange rate passthrough estimates range between 0.11 for the bilateral exchange rate and 0.26 for NEER over a 4-quarter period for the entire SVAR sample.
  - Effect peaks at the fifth quarter for a NEER shock and at the third quarter for a bilateral exchange rate shock; estimates significant only for the NEER shock at the third quarter (95 percent confidence interval excludes zero).
- Variance decomposition:
  - A large part of inflation’s variance is explained by domestic shocks; inflation inertia explains most variation.
  - Domestic policy rate shock is another important domestic factor.
  - Among external shocks, global oil price shocks and exchange rate movements are important.
- Overall headline finding:
  - Combining results, exchange rate passthrough to inflation in Kenya generally range between 0.2 and 0.3 over a period of one year.

### Time variation, robustness, and asymmetries
- Passthrough appears to have weakened in more recent periods:
  - Rolling regressions (10-year windows) show passthrough peaked around 2010 and strengthened temporarily in 2019–20.
  - Bilateral exchange rate equations show renewed strengthening of passthrough in more recent quarters (statistically significant), while recent NEER passthrough estimates are not statistically significant.
- Shorter sample (from 2009) results:
  - Single equation: short-run passthrough generally lower for NEER changes; long-run passthrough larger by about 0.1 (both exchange rate measures) when including short-term interest rate.
  - SVAR: cumulative effect after four quarters about 0.19 for both NEER and bilateral exchange rate (marginally significant at 90 percent).
- Asymmetry:
  - Single equation model with NEER finds passthrough slightly higher for depreciation than appreciation: NEER depreciation passthrough 0.12 versus NEER appreciation 0.10 (specification (3) in Technical Annex Table 2).
  - This asymmetry not found for the bilateral exchange rate specification.

### Monetary policy, expectations, and implications
- CBK institutional developments:
  - Inaugural Monetary Policy Committee meeting in 2008.
  - CBK Act amended in 2012 making price stability a primary objective.
  - Inflation target has come down from 9   percent in FY2011/12 to 5 percent (with ±2.5 percent band).
  - In August 2023 CBK introduced an interest rate corridor around the policy rate (±2.5 percent band) for the overnight interbank rate and launched a Centralized Securities Depository.
- Role of monetary policy:
  - Improved management of inflationary expectations likely contributed to lower inflation volatility and lower passthrough over the past decade.
  - SVAR results: monetary policy has been somewhat more responsive to inflation shocks since 2009; 4-quarter change in policy rate following a 1 percentage point inflation shock is larger for the more recent sample.
  - Variance decomposition shows inflationary developments explain a higher share of variation in the policy rate in recent years; exchange rate shocks also influence policy rate but less than inflation.

### Conclusion and policy takeaway
- Exchange rate passthrough is an important factor for Kenya’s inflation, with combined estimates generally between 0.2 and 0.3 over one year.
- Supply-side and weather-related shocks (e.g., domestic food inflation due to drought) contribute to inflation inertia and are not fully captured by the models.
- Monetary policy action, exchange rate dynamics, and global oil prices are important determinants of Kenya’s inflationary process.
- Given evidence of somewhat higher passthrough to inflation from exchange rate depreciation than appreciation, the current context of exchange rate depreciation calls for monetary policy to remain proactive in anchoring inflationary expectations.

*Italic source: BENCHMARKING TAX PERFORMANCE IN KENYA (IMF); IMF staff calculations on news clips from the International Bureau of Fiscal Documentation (IBFD); annex tables and PSBS material excerpted from the provided PDF content unit.*

### 1. Tax-to -GDP Ratio __________________________________________________________________________________________ 5

### 1. Tax-to -GDP Ratio

### Overview and purpose
- This note examines tax policy and administrative changes in Eastern African Community (EAC) countries to benchmark Kenya’s experience and draw lessons for future tax reforms.
- Uses granular data from a new IMF database on tax measures announced during 1988–2022.
- Main high-level conclusion: EAC policymakers frequently changed tax systems and administrations via tax packages that typically combined measures to narrow tax bases (e.g., exemptions, deductions) and to strengthen tax administrative practices (e.g., electronic payments, tax compliance strategy).
- Caveat noted: frequency of tax measures is not an indicator of the actual revenue impact of such measures; it helps identify reform episodes, their duration, and comprehensiveness.

### Kenya: tax-to-GDP trajectory and composition
- Kenya’s tax-to-GDP ratio trend:
  - Peak at 15.5 percent of GDP in 2014.
  - Fell to 13.1 percent of GDP in 2020.
  - Expected to reach 14.4 percent of GDP in 2023.
- Major compositional shifts:
  - Income taxes declined from 8.0 percent of GDP in 2014 to 6.5 percent of GDP in 2020.
  - Tax ratio initially fell during the pandemic (2020–22) in part due to tax breaks, then began rebounding as COVID-19 measures were reversed on January 1, 2022.
  - The 2023 Finance Act is reported to introduce about 1.5 percent of GDP in new tax policy and administrative measures.

### Comparison with other EAC countries
- Kenya is the only EAC country that experienced a protracted fall in its tax-to-GDP ratio over the last decade.
- Except for South Sudan and the Democratic Republic of Congo (oil exporters), other EAC countries have shown either increasing or relatively stable tax-to-GDP ratios over the same period.
- Kenya moved from being the highest among EAC countries in 2012–15 (average values) to being significantly below Rwanda and Burundi in recent years.

### Database, classifications, and coverage
- Database details:
  - Covers announced and adopted tax policy and administrative changes in EAC economies over 1988–2022.
  - Source: news clips prepared by tax experts from the International Bureau of Fiscal Documentation.
  - Documents direction of changes (INCREASE/DECREASE) in: RATE, BASE, and ADMIN for seven taxes: personal income tax (PIT), corporate income tax (CIT), value added and sale taxes (VAT), social security contributions and payroll taxes (SSC), excise (EXE), trade taxes (TRADE), and property taxes (PRO).
  - Records announcement and implementation dates (mm/dd/yyyy), category of change (e.g., top rate, exemption, tax compliance), whether part of a broader package, and quantitative information on announced size of most rate changes (generally in percentage points).
- Coverage notes:
  - Database excludes South Sudan due to data limitations and has very limited data coverage for Burundi (2012–17).

### Frequency and composition of tax changes (1988–2022, EAC aggregate)
- Total announced changes in EAC countries: 1,845 changes (average about 13 tax policy and administrative changes per year).
- Composition of changes (share of total identified changes):
  - BASE changes: about 43 percent of total identified changes.
  - ADMIN changes: about 36 percent of total identified changes.
  - RATE changes: remaining portion (about 21.7 percent in table summary).
- Directional patterns:
  - BASE changes: about 60 percent were base-narrowing measures.
  - ADMIN changes: about 80 percent intended to strengthen current practices.
  - RATE changes: announcements of rate hikes were equally frequent as rate reductions.
  - Overall, tax policy measures (RATE and/or BASE) represented about 65 percent of total changes and typically entailed a reduction in taxpayers' liabilities (56 percent of total tax policy changes).

### Kenya-specific frequency and composition (1988–2022)
- Kenya announced 594 tax policy and administrative changes (average about 18 changes a year).
- Distribution in Kenya:
  - BASE changes: 270 observations (28.2 percent decrease/loosen; 13.0 percent increase/strengthen; total share 41.2 percent).
  - ADMIN changes: 196 observations (5.9 percent decrease/loosen; 23.9 percent increase/strengthen; total share 29.9 percent).
  - RATE changes: 128 observations (9.6 percent decrease/loosen; 9.9 percent increase/strengthen; total share 19.5 percent).
  - Total Kenya: 594 observations (43.8 percent decrease/loosen; 46.8 percent increase/strengthen; combined frequency share 90.5 percent in table).
- Notable Kenya features:
  - Kenya was the only country in the sample where base-narrowing measures were announced more frequently than measures to strengthen administrative practices (31 percent versus 26 percent of total tax measures announced in the country).
  - Kenya was the only country where the frequency of tax policy changes introducing a reduction in taxpayers' liabilities exceeded 60 percent of total tax policy changes (comparison: 52 percent for EAC excluding Kenya).
  - Kenya was most active in announcing changes to CIT and VAT and also most active in announcing changes to PIT, TRADE taxes, and SSC—most of these being BASE changes.

### Cross-country and tax-type patterns
- Aggregate EAC patterns by tax type (share of total tax measures):
  - CIT: 24 percent of total tax measures.
  - VAT: 20 percent of total tax measures.
  - Excise (EXE): 16 percent of total tax measures.
- Nature of changes by tax:
  - CIT and VAT: about half of measures consisted of tax base changes; base-narrowing accounted for the bulk (CIT base-narrowing 60 percent; VAT base-narrowing 68 percent).
  - Excise: 49 percent of excise measures entailed rate changes; excise rate measures were predominantly hikes (75 percent of total excise rate changes).
  - Administrative changes for CIT and VAT accounted for a bit more than one-third of total changes for those taxes; administrative changes for excise were relatively infrequent (13 percent of total excise changes).
  - Measures to strengthen administrative practices were the most common within each tax type (average frequency above 70 percent of respective administrative changes).
- Country differences (high-level):
  - Kenya: frequent base-narrowing for PIT, CIT, VAT, and TRADE; net frequency of base-narrowing exceeded broadening for those taxes.
  - Uganda: most changes to EXCISE and measures to adjust administration of the ensemble of taxes (TAX); EXCISE measures in Uganda appeared to have almost equal shares of RATE and BASE changes.
  - Tanzania: most positive net frequency for EXCISE measures and most negative net frequency for RATE changes for PIT, CIT, VAT, and TRADE.

### Implications, interpretation, and limitations
- Interpretation offered:
  - Frequent introduction of base-narrowing measures, if sustained over time, can result in an erosion of the tax base and help explain declines in tax-to-GDP ratios (as observed in Kenya since 2014).
  - Policymakers often combined measures to realign rates across taxes, permit some base-narrowing (exemptions/deductions), and improve administrative practices.
- Limitations stressed:
  - Frequency of tax measures does not indicate the magnitude of revenue impact; the database captures counts and directions, not revenue elasticity or fiscal magnitude.
  - Tax outturns are also shaped by macro conditions and behavioral responses, not only by tax decisions.

*Source: BENCHMARKING TAX PERFORMANCE IN KENYA (IMF), section "1. Tax-to -GDP Ratio".*

### 9.      Most tax changes were announced as part of a broad package of measures that

### 1kenea2024002 - 9.      Most tax changes were announced as part of a broad package of measures that

### Overview: prevalence and composition of tax packages (1988–2022)
- During 1988–2022, EAC countries announced virtually all tax changes (94 percent of total) as part of a broad package of tax measures.
- The scope of the average tax package was typically broad: almost 2 out of three packages in the sample included at least one tax policy measure (tax rate and/or tax base) and one administrative change.
- Only in 25 percent of the cases were tax rates, tax bases, or administrative measures announced alone.
- The finding implies that assessments focusing on an individual tax change (for example, a change in the top rate of PIT) will likely be severely biased because policymakers typically changed other tax rates and/or tax bases and/or administrative practices at the same time.

### Co-occurrence of increases and decreases; use of offsets
- Tax packages frequently entailed changes in opposite directions, indicating routine use of offsets:
  - Packages that included measures aimed at lowering tax liabilities (lowering tax rates and/or narrowing tax bases and/or loosening administrative practices) were more often accompanied by at least one measure to increase tax liabilities (increase tax rate and/or broaden tax base and/or strengthen administrative procedure) — this pattern occurred in 88 percent of total tax packages.
  - Packages with lowering/loosening measures but no accompanying increases/broadening/strengthening represented 12 percent of total packages.
- EAC countries more often announced packages consisting only of administrative measures to strengthen existing practices without offsetting tax policy measures.
- Implication: evaluating the revenue impact of a specific tax change requires accounting for simultaneous, potentially offsetting changes in other tax policy and administrative dimensions.

### Time variation and country differences
- Policymakers in EAC countries became more active in changing taxes and/or tax administrations after the Global Financial Crisis.
- During the pandemic (2020–22), most countries further intensified interventions to cushion vulnerable households and businesses and, in some cases, later repealed previously introduced tax changes (example given: Kenya).
- Kenya:
  - Featured the highest number of announcements of tax changes for many years after 2014.
  - Experienced a downward trend in the tax-to-GDP ratio during 2015–22, suggesting a need to investigate whether the announcements tended to reduce taxpayer liabilities.
  - Showed a pattern (since 2009 except 2012 and 2020) of more measures to strengthen tax administrative practices (positive net frequency) combined with more measures to reduce tax rates and/or narrow tax bases (negative net frequency of tax policy measures).
- Uganda:
  - Increased frequency of tax changes significantly from 2017 onwards.
  - Over the same period, tax-to-GDP appeared to improve significantly.
  - Displayed at least six years of co-occurrences of more measures to strengthen tax administrative procedures and raise tax policy (positive net frequencies).
- Cross-country implication: differing tax performance (e.g., between Kenya and Uganda) may relate to Kenya’s more pronounced tendency to frequently announce tax policy measures that reduce tax rates and narrow tax bases even while strengthening administration.

### Interaction with IMF programs and program-related measures
- EAC countries generally announced more measures to strengthen than relax taxes covered by program conditionality, often accompanied by additional changes in other taxes.
- Example: Kenya introduced a minimum alternative tax and a digital tax in January 2021 (both classified as base-broadening measures and discussed in program negotiations). After approval of the EFF/ECF program on April 2, 2021, authorities:
  - Introduced an exemption from the minimum alternative tax for an air company with a government share above 45 percent and its subsidiaries, and
  - Limited the application of the digital tax to non-residents.
  - These later design changes are captured in the database as CIT base-narrowing measures.
- Another example: VAT base-broadening measures adopted in early 2021 were accompanied in January 2022 by the introduction of several exemptions for medical supplies amid a third wave of COVID-19; these are classified in the database as base-narrowing measures.
- Note: a negative net frequency (more relaxations than strengthenings) does not necessarily imply an overall negative revenue impact; revenue impact requires detailed assessment.

### Key policy implication and recommendation for Kenya
- Kenya needs to strengthen tax collection consistent with authorities’ objectives of sustained increases in tax revenues to meet their development agenda.
- A key milestone is the timely adoption of Kenya’s first Medium-Term Revenue Strategy (SR¶17), developed with IMF support.
- The Strategy aims to increase revenues by 5 percentage points of GDP by FY2026/27 through measures that broaden the tax base and strengthen tax compliance.

### Data coverage (selected)
- Dataset period referenced throughout analysis: 1988–2022.
- Country data coverage noted in Appendix I:
  - Kenya 1988-2022
  - Burundi 1989-1995 ; 2004-2017
  - Congo, Democratic Republic of 2001 ; 2009-2022
  - Rwanda 1999-2000 ; 2007-2022
  - Tanzania 1988-2022
  - Uganda 1988-2022

*Source: IMF, staff calculations on news clips from the International Bureau of Fiscal Documentation (IBFD).*

### Appendix I. Table 4. Kenya: Tax Policy and Administrative Changes Announced, 1988–2022

### Appendix I. Table 4. Kenya: Tax Policy and Administrative Changes Announced, 1988–2022

### Tax policy and administrative changes (table highlights)
- The source presents counts of observations, count of country years, and average number of measures by tax type and change category for PIT, CIT, VAT, SSC, and EXE. Example entries (preserving source formatting and numeric values exactly):
  - PIT (top section)
    - Summary line: 99283.535201.864262.5
    - RATE: 1181.4431.3771.0
    - Statutory rates: 111.0111.000-
    - Top rate: 761.2111.0661.0
    - Surcharges: 221.0111.0111.0
    - BASE: 62252.51491.648251.9
    - Standard relief: 12101.2111.011101.1
    - Capital gains (base): 761.2221.0541.3
    - ADMIN: 26141.917111.5971.3
    - Integrity taxpayer base: 441.0441.000-
    - Payments: 15111.41181.4441.0
    - Other (ADMIN): 541.3111.0441.0
  - CIT (middle of table)
    - Summary line: 147314.754163.493293.2
    - RATE: 23171.4541.318161.1
    - Statutory rates: 441.000-441.0
    - Top rate: 871.1111.0771.0
    - Surcharges: 221.0111.0111.0
    - BASE: 78272.920131.558232.5
    - Investment promotion: 21171.2221.019171.1
    - Loss-carry rules: 551.0111.0441.0
    - Thin capitalization: 531.7111.0431.3
    - Other base changes: 34181.91091.124141.7
    - ADMIN: 46192.429112.617141.2
    - Integrity taxpayer base: 971.3651.2331.0
    - Payments: 20141.41081.310101.0
    - Other (ADMIN): 1181.4751.4441.0
  - VAT (continued)
    - Summary line: 137284.971262.766252.6
    - RATE: 28181.612101.216141.1
    - Standard rate: 971.3441.0551.0
    - BASE: 65232.821141.544202.2
    - Exemptions on food: 871.1221.0651.2
    - ADMIN: 44222.038221.7632.0
    - Payments: 13101.31191.2221.0
    - Other (ADMIN): 771.0661.0111.0
  - SSC (social security contributions)
    - Summary line: 1061.7961.5111.0
    - RATE: 321.5321.500-
    - Employee: 221.0221.000-
    - BASE: 522.5422.0111.0
    - Employee (base): 321.5221.0111.0
    - ADMIN: 221.0221.000-
  - EXE (excises)
    - Summary line: 83243.558212.825151.7
    - RATE: 35172.127151.8871.1
    - Alcohol products: 991.0881.0111.0
    - Tobacco: 661.0441.0221.0
    - Oil products: 541.3441.0111.0
    - BASE: 30112.71691.814101.4
    - Alcohol products (base): 441.0111.0331.0
    - Tobacco (base): 321.5321.500-
    - ADMIN: 18111.61591.7331.0
    - Integrity taxpayer base: 331.0221.0111.0
    - Payments: 331.0221.0111.0
    - Other (ADMIN): 761.2651.2111.0
- The table distinguishes Decrease / Total changes / Increase but retains detailed cell-level numeric entries as presented above.

### Public sector balance sheet (PSBS): scope, findings, and key metrics
- Legal and institutional context
  - Public Financial Management Act (PFMA) 2012, amended in 2023, mandates the National Treasury (NT) to be the custodian of an inventory of national government assets and to manage the level and composition of public debt, guarantees and other financial obligations.
  - NT actions: consolidated financial statements for almost all public sector entities; establishment of a Fiscal Risk Committee and a fiscal risk register; ongoing implementation of a treasury single account and accrual-based IPSAS; NT estimated public sector size and composition in 2014 and 2018.
- PSBS headline metrics (consolidated, FY2022/23)
  - Total public sector assets: 101 percent of GDP.
  - Public sector estimated financial assets: 14 percent of GDP.
  - Public sector estimated financial liabilities: 131 percent of GDP.
  - Net financial worth: negative 117 percent of GDP.
  - Net worth: minus 30 percent of GDP.
- Fiscal deviations and risks
  - Public debt consistently exceeded medium-term projections with an annual rate of about 4 percent of GDP from 2014 to 2022.
  - Contingent liabilities from counties and state-owned corporations contributed materially to deviations.
- Public corporations, SAGAs, and SCs performance and fiscal implications
  - Public corporations, SAGAs and SCs collectively hold assets worth of 67 percent of GDP as of end-June 2023.
  - Commercial corporations are estimated to manage half of these assets.
  - Dividends and taxes paid by SCs (percent of GDP, FY2015/16–FY2022/23 averages and series as reported):
    - Investment Income (CBK): 0.00 0.00 0.01 0.11 0.07 0.04 0.03 0.04
    - Investment Income - Others: 0.36 0.27 0.27 0.33 0.33 0.25 0.26 0.30
    - Taxes paid: n.a. n.a. n.a. n.a. 0.24 0.13 0.13 0.18
    - Total (last printed row fragment): 0.64 0.42 0.42 0.52
  - On average, dividends from these entities averaged 0.3 percent of GDP from FY2015/16 to FY2022/23, with a minimal contribution through taxes at 0.13 percent of GDP in FY2022/23.
  - Dividends remitted by CBK and profitable SCs totaled 0.29 percent of GDP (Table 1 context).
  - SAGAs/SCs losses and equity
    - In FY2022/23, 242 SAGAs/SCs incurred losses amounting to 0.7 percent of GDP, up from 183 entities with losses equivalent to 0.5 percent of GDP in FY2021/22.
    - As of June 30, 2023, there were 526 SAGAs/SCs, from 500 entities as of end-June 2022, of which 16 have not been included in the consolidated financial statements (11 in FY2021/22).
    - In FY2022/23, 18 SAGAs/SCs reported negative equity, totaling 1.5 percent of GDP, up from 1.2 percent of GDP in FY2021/22.
  - Implication: dividends and CBK profits fall short of covering SAGAs/SCs losses; budget support or accumulation of payables/non-equity liabilities may follow.
- Compensation and fiscal coverage insights
  - National government compensation of employees: declined from 4.3 percent of GDP in FY2019/20 to 3.8 percent of GDP in FY2022/23.
  - SAGAs and SCs compensation of employees: increased by 0.6 percent of GDP over the same period.
  - Entire public sector compensation of employees: remained at about 8.4 percent of GDP.
  - General government compensation of employees estimated at 7 percent of GDP in FY2022/23 or constituting 49 percent of tax revenues.
- Potential gains from improved asset management
  - International experience indicates revenue gains from improved management of non-financial public corporations and government financial assets could potentially reach to 3 percent of GDP annually (IMF, 2018).

### Recommendations and implications (as presented)
- Broaden fiscal analysis from national government flows to full public sector stocks and flows to better assess fiscal sustainability and risks.
- Use the PSBS to:
  - Improve transparency around assets and liabilities not currently in fiscal reports (natural resources, pension liabilities, PPP contracts, other claims and payables).
  - Strengthen asset management to boost returns, taxes, and dividends from public corporations.
  - Monitor and manage contingent liabilities from counties and state-owned enterprises to reduce unexpected fiscal pressures.
- Continue PFM reforms: consolidate financial statements, implement accrual IPSAS, maintain the Fiscal Risk Committee and fiscal risk register, and complete treasury single account implementation to improve fiscal risk identification and management.

*Source: IMF, staff calculations on news clips from the International Bureau of Fiscal Documentation (IBFD); excerpted content from the provided PDF content unit.*

### 11.      The strength of a country’s PSBS matters for both macroeconomic stability and

### 11.      The strength of a country’s PSBS matters for both macroeconomic stability and

### Importance of Public Sector Balance Sheets (PSBS)
- Economies with robust public sector balance sheets experience shallower recessions and tend to recover faster after economic downturns.
- Stronger balance sheets provide governments greater flexibility to employ countercyclical policies, such as increasing spending during economic downturns.
- Financial markets account for government assets and net (financial) worth when pricing sovereign bonds (Yousefi (2009)).
- Expanding analysis to assets and non-debt liabilities captures fiscal operations conducted outside the national government, particularly by SAGAs, social funds, state corporations, and PPP contracts.
- Excluding non-debt liabilities such as pending bills, unpaid tax refunds, and legal claims weakens assessment of public finances; these non-debt liabilities effectively act as zero-yield assets (inflation adjusted, negative yield) for the private sector.

### Methodology and Coverage for Kenya’s PSBS
- PSBS compilation is data-intensive and follows GFSM 2014 reporting of accrual information and balance sheets.
- Kenya’s PSBS scope (static PSBS) includes the national government (BCG units), SAGAs, counties (local governments), social securities, and state corporations (public financial and non-financial corporations).
- Intertemporal PSBS combines discounted future flows of revenues and expenditures with the static balance sheet; Kenya’s intertemporal PSBS is not estimated in the paper.
- Kenya does not produce a formal PSBS, but consolidated financial statements for different public sector perimeters enable estimation of a static PSBS.
- The estimated PSBS FY2022/23 uses information from fiscal reports and prior IMF capacity development missions and other studies.

### Key Statistics and Composition: PSBS FY2022/23
- Estimated net worth (June 2023): minus 30 percent of GDP.
- Public sector assets: 101 percent of GDP.
  - Non-financial assets: 87 percent of GDP (infrastructure, buildings, public land holdings, fixed assets and equipment held by SAGAs and state corporations).
  - Financial assets: 14 percent of GDP, composed of:
    - Cash and deposits: 3.1 percent of GDP
    - Debt securities: 1 percent of GDP
    - Equity investment: 2 percent of GDP
    - Receivables: 7.9 percent of GDP
- Public sector liabilities: 131 percent of GDP, composed of:
  - Government debt securities and loans: 70 percent of GDP
  - Debt securities from state corporations (SCs): 6 percent of GDP
  - Currency and deposits owed by the CBK and financial corporations: 6 percent of GDP
  - Actuarial pension obligations: 33 percent of GDP
  - Pending bills and other payables, including PPP contracts: 16 percent of GDP
- Cross-holdings across public sector segments: approximately 44 percent of GDP, including:
  - Government equity claims on SAGAs/SCs: 35.6 percent of GDP
  - Government and SAGAs/SCs deposits at the CBK: 4.8 percent of GDP
  - Government securities held by SCs and the CBK: 1.7 percent of GDP
  - Receivables/payables: 1.4 percent of GDP
- Foreign exchange exposure:
  - Foreign exchange-denominated assets: about 5 percent of GDP
  - Foreign exchange liabilities: about 40 percent of GDP

### Comparative and Vulnerability Findings
- Total liabilities of 131 percent of GDP place Kenya on par with peers such as South Africa and Senegal and comparable with many countries that have an estimated PSBS.
- Kenya’s financial assets (14 percent of GDP) are much smaller than its debt levels (70 percent of GDP) and total liabilities (131 percent of GDP), leaving limited liquid assets to meet gross financing needs.
- Low asset holdings relative to liabilities make Kenya vulnerable to external shocks; asset valuations (e.g., natural resources) can be cyclical and illiquid when financing needs are most pressing.
- The PSBS omits mineral and energy resources valuation due to limited presence; Kenya has almost nonexistent nonrenewable natural resources per cited data.
- Cross-holdings do not change net PSBS totals but can transmit risks between sectors, potentially affecting the entire public sector.

### Trends, Drivers, and Recent Evolution
- Public sector net worth deteriorated by 25 percentage points of GDP compared with FY2017/18.
  - Drivers include reduction of infrastructure due to lower investment to offset amortization, weakening financial performance of SAGAs and SCs, and liabilities increasing by 10 percentage points of GDP from higher public debt, non-equity liabilities of SAGAs and SCs, and pending bills of national government and counties.
- Fiscal flows FY2022/23 (consolidated indicators from table):
  - Revenue: 24.0 percent of GDP (components: National Government 16.7; Counties 3.2; SAGAs/SCs 10.1; cross-holdings -6.0 per table structure)
  - Expenditure: 28.4 percent of GDP (components: National Government 22.6; Counties 3.1; SAGAs/SCs 8.8; cross-holdings -6.0)
  - Net/Lending Borrowing (Profit/Loss): -4.4 percent of GDP (components: National Government -5.9; Counties 0.1; SAGAs/SCs 1.4; cross-holdings 0.0)

### Limitations of Balance Sheet Analysis
- Data quality depends on public financial management systems and adherence to accounting principles; public corporations’ reliability depends on external audits.
- Valuation challenges exist, particularly for nonfinancial assets that are not traded.
- The public sector comprises diverse entities with distinct constraints and risks, requiring entity-specific analysis.
- Recognition of assets does not remove vulnerabilities associated with high public debt because many assets are illiquid and asset valuations can fall in downturns.

### Policy Implications and Recommendations
- Focusing on balance sheet indicators (net financial and net worth) provides a more comprehensive fiscal assessment than debt and deficit alone.
- Implementing the medium-term tax strategy (MTRS) is expected to generate additional revenue equivalent to about 5 percent of GDP, enabling:
  - Strengthening the government balance sheet
  - Less reliance on debt-financing
  - Increased spending for social policies
- Clearing government arrears, particularly pending bills, will strengthen PSBS in the medium term:
  - Settling arrears prevents penalties and strengthens SAGAs and SCs balance sheets.
  - If settled through borrowing or securitization, short-term balance sheet effect is neutral; settling via increased tax revenue or spending cuts would immediately improve financial net worth.
- Pension reform (January 2021) converting defined-benefit schemes to defined-contributory is projected to decrease national government pension obligations by about 1 percent of GPD annually for the next 30 to 35 years, but requires:
  - Proper monitoring, governance, periodic evaluation, and mitigation of potential risks, especially given existing pension arrears.
  - As of end-June 2023, SAGAs and SCs had pension arrears of KSh.46.8 billion, equivalent to 55 percent of the total assets of the new contributory scheme or 12 percent of the combined total assets of the National Social Security Fund and the Public Service Superannuation Fund (new scheme).
- Expand and centralize PPP risk management:
  - PPPs can expand PSBS; their impact on net worth depends on asset productivity and containment of fiscal risks.
  - PPPs may create illusion of fiscal space; they carry explicit or implicit contingent liabilities from asymmetric information and complex project risks.
  - A centralized framework integrating PPP projects into the national public investment and budget framework is essential; considering limits on PPP portfolio size is an option Kenya could consider.

*Sources: Kenyan National Treasury; IMF, Government Finance Statistics database; and IMF staff calculations.*

### 29.      Reforms aimed to rationalize SAGAs, improve the governance of SCs, and privatize

### Reforms aimed to rationalize SAGAs, improve the governance of SCs, and privatize selected public entities will strengthen Kenya balance sheet

### Reforms and expected fiscal effects
- The number of SAGAs and SCs rose from 350 in June 2020 to 526 in June 2023, primarily due to the expansion of vocational education and training colleges.
- Authorities are reviewing and rationalizing SAGAs/SCs in line with service delivery mandates.
- A new ownership policy for government owned enterprises will enable an enhanced governance framework for commercial SCs to improve service delivery and profitability.
- A strategic privatization program is anticipated to:
  - generate revenue streams for the government;
  - curtail transfers to non-profitable entities; and
  - alleviate the overall cost of capital borne by public sector assets.
- Public sector assets incur continuous costs (the cost of capital) including borrowing expenses, tax revenues for capital acquisition, maintenance costs, and potential expenses when assets fail to meet expectations.
- The reform agenda creates an opportunity to implement distinct governance and monitoring approaches for:
  - public entities funded by the budget and extra-budgetary units, and
  - entities operating on a commercial basis.

### Caveats and data limitations
- The PSBS is based on published statistics and a few estimates, using consolidated financial statements of MDAs, counties, and SAGAs/SCs for FY2022/23 and other official sources.
- Fixed assets of MDAs have been estimated for FY2022/23; authorities believe the value of fixed assets is higher than the PSBS estimates, and the non-inclusion of natural resources underestimates fixed assets.
- Cross-holdings of assets and liabilities are likely higher than disclosed because financial statements lack detailed disclosure; this does not affect financial and net worth aggregates.
- Liabilities are estimated from national government debt, pending bills, and SAGAs/SCs liabilities reported by the NT and presented in consolidated financial statements.
- Liabilities related to PPP contracts use the World Bank database but are presented at half of their stock based on discussions with government officials.
- Pension obligations:
  - Two pension schemes administered by the government: one non-contributory (defined benefits) and one contributory scheme introduced in January 2021.
  - The most recent actuarial valuation available for the non-contributory scheme is from a World Bank study in 2016, estimating pension liabilities at 30 percent of GDP (IMF, 2020).
  - The 2020 Kenya Fiscal Transparency Evaluation Update included an actuarial obligation to the social security sector of 3.3 percent of GDP in FY2017/18 PSBS; due to lack of data, an estimated value of 3.3 percent of GDP is included in the PSBS.
  - Some SCs manage their own defined benefit schemes, typically presented in net values (value of assets minus actuarial obligations). Example: CBK reports a net asset of Ksh.5.0 billion against a fair value of scheme assets of Ksh.29.8 billion as of end-June 2023.
  - Some state corporations have closed defined benefit schemes and moved to contribution schemes (Kenya Power and Lighting Company closed its defined benefit scheme in June 2006; Kenya Electricity Generating Company closed its scheme in December 2011).

### Key findings from the FY2022/23 PSBS and fiscal implications
- SAGAs and SCs manage assets worth about 70 percent of GDP, yet only a few contribute to the national budget.
- Nearly half of these entities operated at a loss over the last two fiscal years, amounting to over 1 percent of GDP or approximately half of the revenues from the value-added tax on domestic goods and services in FY2022/23.
- Improvement in national government fiscal indicators has been partially achieved at the expense of the rest of the public sector, reflected in increased wage bill expenses and accumulation of pending bills in SAGAs and SCs.
- The PSBS reveals significant non-debt liabilities accumulated over the years, requiring improved asset management and policy measures to contain or improve the net financial worth position.
- Enhanced transparency from the PSBS provides policymakers with insights to formulate effective fiscal policies that can contribute to higher economic growth and improved fiscal space.

### Recommendations and PFM priorities
- Categorize SAGAs and SCs into portfolios (policy and service delivery versus commercial operations) and differentiate them in financial and non-financial terms, with special focus on social sector entities and those with pension obligations.
- Conduct an actuarial evaluation for the defined-contribution scheme and monitor the new scheme rigorously.
- Strengthen the PFM system to address:
  - accumulation of pending bills;
  - delays in tax refunds; and
  - lack of budget credibility.
- Specific PFM measures recommended:
  - prepare realistic budgets;
  - introduce multi-year commitments for investment projects;
  - execute budgets in accordance with parliamentary appropriations;
  - enhance digitalization of PFM systems; and
  - strengthen procurement processes.
- These measures are expected to form integral components of the new PFM reform strategy being developed by the NT.
- A credible PFM system is essential to support the debt anchor set at 55 percent of GDP in present value and to prevent accumulation of non-debt liabilities.
- Fiscal statistics should at least encompass the central government and provide comprehensive reporting on public sector liabilities.

### HM Treasury (United Kingdom) Balance Sheet Framework (Annex I) — principles and application
- Balance Sheet Review (BSR) objectives: dispose of assets that no longer serve a policy purpose, improve returns on retained assets, reduce risk and cost of liabilities, release resources for investment, and improve sustainability.
- HM Treasury balance sheet management principles:
  - secure maximum value for taxpayers from the government’s assets and liabilities;
  - enhance transparency over balance sheet decisions;
  - optimize management and mitigation of balance sheet risks;
  - safeguard overall public sector net worth; and
  - strengthen fiscal sustainability.
- The framework divides public sector assets and liabilities into three portfolios: policy, financial, and commercial, with distinct management objectives, governance arrangements, and exit strategies.
- The framework is used to:
  - update central guidance for risk management and optimizing returns;
  - evaluate cases for significant asset sales and balance sheet transactions;
  - inform how credit risk should be managed across portfolios;
  - inform mandates of institutional vehicles delivering policy priorities;
  - identify economies of scale within portfolios; and
  - consider developing an investment strategy for future performance expectations.

### Methodology notes and selected PSBS metrics (Annex II and Table 1)
- Main data sources: National Treasury, Office of the Controller of Budget, Kenya Bureau of Statistics, Central Bank of Kenya, IMF Government Financial Statistics Database, IMF TA Reports, World Bank database on PPP, The Changing Wealth of Nations 2021.
- Few items estimated:
  - Non-financial assets for MDAs and Counties (infrastructure assets estimated using FTE FY2017/18 plus NT fixed asset transactions; amortization rate of 2 percent annually applied).
  - Pension obligations (non-contributory scheme) assumed the same as FTE 2020: 30 percent of GDP for the national government and 3.3 percent for SAGAs/SCs.
  - Discounted at 5 percent, pension expense projections stock in FY2020/21 was 16.6 percent of GDP.
  - PPP portfolio based on World Bank database but reduced by 50 percent, estimated at 2.2 percent of GDP.
- Selected PSBS FY2022/23 public sector aggregates (percent of GDP):
  - Entities manage assets worth about 70 percent of GDP.
  - Insurance, pension, and standardized guarantee schemes (liabilities): 33.3 percent of GDP (of which, pension liabilities 33.3 percent of GDP).
  - PPP liabilities: 2.2 percent of GDP.
  - Net financial worth (Public Sector): -117.1 (percent of GDP).
  - Net worth (Public Sector): -30.0 (percent of GDP).
  - Flows (Public Sector):
    - Revenue: 24.0 (percent of GDP).
    - Expenditure: 28.4 (percent of GDP).
    - Net/Lending Borrowing (Profit/Loss): -4.4 (percent of GDP).

*Italic source: IMF staff summary of the PSBS material contained in the original document.*

### 2.      Past episodes of large exchange rate

### 2.      Past episodes of large exchange rate

### Historical context and inflation patterns
- Past episodes of large exchange rate depreciations often coincided with higher inflation but not always; food prices (notably maize) and droughts have driven some past inflation episodes.
- A recent example of muted inflationary response to a noticeable depreciation was 2020, when the economic slowdown following the COVID-19 pandemic dominated inflationary developments.
- Given food prices’ importance in Kenya’s CPI, weather-related supply shocks can cause inflationary inertia not fully attributable to exchange rate movements.

### Literature and prior estimates
- Early Kenya studies: Ryan and Milne (1994) report importance of exchange rate movements and oil prices; Durevall and Ndung’u (1999) identify exchange rate, foreign price level, and terms of trade as “proximate determinants”.
- Cross-country and regional findings:
  - Choudhri and Hakura (2006): passthrough ranges from 0.09 on impact to 0.35 after four quarters and 0.38 after twenty quarters.
  - Revelli (2020): passthrough varies between 0.18 and 0.58 over one year (single equation), lower peak of 0.3125 using VAR.
  - April 2023 IMF Regional Economic Outlook for Sub-Saharan Africa: one-year passthrough at 0.22–0.25 for the region; for non-pegged countries passthrough estimated at 0.28.
  - Razafimahefa (2012): passthrough 0.4 for the region.
  - Caselli and Roitman (2016): annual passthrough 0.22 in emerging economies.

### Estimation approach and data
- Two approaches used on quarterly data between 1995 and 2023Q2:
  - Single equation model (local-projections style), including Kenyan inflation, global maize inflation, crude oil prices, changes in nominal exchange rate, and changes in short-term interest rate (three-month T-bill rate). Changes measured on y/y basis to use stationary variables.
  - Non-recursive SVAR model with two blocks: exogenous external variables (global oil prices, U.S. real GDP, U.S. federal funds rate) and Kenyan domestic block (consumer prices, real GDP, central bank policy rate, three-month T-bill rate, nominal exchange rate). Variables estimated in first differences with two lags; external variables treated as exogenous.
- Data notes:
  - Estimations use quarterly data 1995–2023Q2; CBK policy rate series from 2003Q3 onward.
  - Exchange rates defined per unit of Kenyan shillings: a positive change is an appreciation of the shilling.

### Main estimation results and key statistics
- Single equation model (y/y changes):
  - On-impact passthrough for a 1 percent change in exchange rate: range between 0.06 and 0.12 (specifications (1), (2), (4), and (5) in Technical Annex Table 2).
  - Cumulative passthrough over a longer period (after accounting for lagged inflation): range between 0.22 and 0.42.
  - Passthrough estimates were found to be higher for NEER under these specifications.
- Non-recursive SVAR (accumulated impulse responses):
  - Exchange rate passthrough estimates range between 0.11 for the bilateral exchange rate and 0.26 for NEER over a 4-quarter period for the entire SVAR sample.
  - Effect peaks at the fifth quarter for a NEER shock and at the third quarter for a bilateral exchange rate shock; estimates significant only for the NEER shock at the third quarter (95 percent confidence interval excludes zero).
- Variance decomposition:
  - A large part of inflation’s variance is explained by domestic shocks; inflation inertia (shocks to inflation itself) explains most variation.
  - Other important domestic factor: domestic policy rate shock.
  - Among external shocks, global oil price shocks and exchange rate movements are important; global oil price shock explains a modestly higher share when bilateral exchange rate is used.
- Overall headline finding:
  - Combining results, exchange rate passthrough to inflation in Kenya generally range between 0.2 and 0.3 over a period of one year.

### Time variation, robustness, and asymmetries
- Passthrough appears to have weakened in more recent periods:
  - Rolling regressions (10-year windows) show passthrough peaked around 2010 and strengthened temporarily in 2019–20.
  - Bilateral exchange rate equations show renewed strengthening of passthrough in more recent quarters (statistically significant), while recent NEER passthrough estimates are not statistically significant.
- Shorter sample (from 2009) results:
  - Single equation: short-run passthrough generally lower for NEER changes; long-run passthrough larger by about 0.1 (both exchange rate measures) when including short-term interest rate, though not all parameters are statistically significant for the shorter sample.
  - SVAR: cumulative effect after four quarters is smaller for a NEER shock but not for a bilateral exchange rate shock; estimated cumulative effect after four quarters about 0.19 for both NEER and bilateral exchange rate (marginally significant at 90 percent).
- Asymmetry between depreciation and appreciation:
  - Single equation model with NEER (modified per Carrière-Swallow et al (2023)) finds passthrough slightly higher for depreciation than appreciation: NEER depreciation passthrough 0.12 versus NEER appreciation 0.10 (see specification (3) in Technical Annex Table 2).
  - This asymmetry not found in specification involving the bilateral exchange rate.

### Monetary policy, expectations, and implications
- Central Bank of Kenya (CBK) institutional developments:
  - Inaugural Monetary Policy Committee meeting in 2008.
  - CBK Act amended in 2012 making price stability a primary objective.
  - Inflation target has come down from 9   percent in FY2011/12 to 5 percent (with ±2.5 percent band).
  - In August 2023 CBK introduced an interest rate corridor around the policy rate (±2.5 percent band) for the overnight interbank rate and launched a Centralized Securities Depository.
- Role of monetary policy:
  - Improved management of inflationary expectations via monetary policy likely contributed to lower inflation volatility and lower passthrough over the past decade.
  - SVAR results: monetary policy has been somewhat more responsive to inflation shocks since 2009; 4-quarter change in policy rate following a 1 percentage point inflation shock is larger for the more recent sample.
  - Variance decomposition shows inflationary developments explain a higher share of variation in the policy rate in recent years; exchange rate shocks also influence policy rate but less than inflation.

### Conclusion and policy takeaway
- Exchange rate passthrough is an important factor for Kenya’s inflation, with combined estimates generally between 0.2 and 0.3 over one year.
- Supply-side and weather-related shocks (e.g., domestic food inflation due to drought) contribute to inflation inertia and are not fully captured by the models.
- Monetary policy action, exchange rate dynamics, and global oil prices are important determinants of Kenya’s inflationary process.
- Continued strengthening of the monetary policy framework and a more responsive monetary policy have helped lower inflation volatility and, at least in part, reduce exchange rate passthrough in the past decade.
- Given evidence of somewhat higher passthrough to inflation from exchange rate depreciation than appreciation, the current context of exchange rate depreciation calls for monetary policy to remain proactive in anchoring inflationary expectations.

*Source: IMF staff estimates and Technical Annex (quarterly data 1995–2023Q2; CBK policy rate series from 2003Q3).*

### Annex I. Table 1. Kenya: Results of Unit Root Testing

### Annex I. Table 1. Kenya: Results of Unit Root Testing

### Unit root test results — t-statistics at level
- Augmented Dickey-Fuller (ADF) test (With constant):
  - lwoil: -1.9051
  - lwmaize: -2.0583
  - lusgdp: -1.6298
  - ffr: -2.5138
  - lcpi: -0.9010
  - lgdp: -0.0061
  - lnneer: -1.3706
  - ler: -1.5944
  - cbkpr: -4.7918***
  - fitb3m: -2.5774
- ADF test (With constant and trend):
  - lwoil: -2.8739
  - lwmaize: -2.9231
  - lusgdp: -2.7886
  - ffr: -2.5089
  - lcpi: -2.2039
  - lgdp: -3.5182**
  - lnneer: -4.1017***
  - ler: -2.1574
  - cbkpr: -4.7985***
  - fitb3m: -2.5028
- ADF test (Without constant and trend):
  - lwoil: -0.3481
  - lwmaize: 0.0546
  - lusgdp: 4.8626
  - ffr: -1.5576
  - lcpi: 6.2317
  - lgdp: 4.4623
  - lnneer: -1.7608*
  - ler: 2.4779
  - cbkpr: -0.3927
  - fitb3m: -1.6198*

- Phillips Perron (PP) test (With constant):
  - lwoil: -1.9674
  - lwmaize: -1.6707
  - lusgdp: -1.9412
  - ffr: -2.0827
  - lcpi: -0.7132
  - lgdp: 0.0999
  - lnneer: -2.2999
  - ler: -1.6190
  - cbkpr: -3.6266***
  - fitb3m: -1.9752
- PP test (With constant and trend):
  - lwoil: -2.3791
  - lwmaize: -2.4454
  - lusgdp: -2.6579
  - ffr: -1.5481
  - lcpi: -2.0526
  - lgdp: -5.9642***
  - lnneer: -4.4085***
  - ler: -3.3964*
  - cbkpr: -3.6056**
  - fitb3m: -2.0484
- PP test (Without constant and trend):
  - lwoil: -0.3792
  - lwmaize: 0.0207
  - lusgdp: 5.6959
  - ffr: -1.5045
  - lcpi: 11.9709
  - lgdp: 12.7616
  - lnneer: -1.8440*
  - ler: 2.5022
  - cbkpr: -0.5218
  - fitb3m: -1.1687

Notes on level results:
- Statistical significance markers shown in the table: ***, **, * denote significance at 1, 5, or 10 percent level respectively. A statistical significance at those levels would not allow accepting the null hypothesis of presence of a unit root.

### Unit root test results — t-statistics at first difference
- ADF test (With constant):
  - lwoil: -8.3048***
  - lwmaize: -7.9924***
  - lusgdp: -12.5908***
  - ffr: -5.8671***
  - lcpi: -8.7479***
  - lgdp: -5.0331***
  - lnneer: -8.8942***
  - ler: -10.4442***
  - cbkpr: -7.2363***
  - fitb3m: -8.7856***
- ADF test (With constant and trend):
  - lwoil: -8.2755***
  - lwmaize: -7.9803***
  - lusgdp: -12.7259***
  - ffr: -6.0256***
  - lcpi: -8.7610***
  - lgdp: -5.0026***
  - lnneer: -8.8084***
  - ler: -10.3693***
  - cbkpr: -7.1965***
  - fitb3m: -8.8106***
- ADF test (Without constant and trend):
  - lwoil: -8.3073***
  - lwmaize: -8.1376***
  - lusgdp: -10.2483***
  - ffr: -5.8935***
  - lcpi: -2.2617**
  - lgdp: -2.1023**
  - lnneer: -8.8265***
  - ler: -10.1396***
  - cbkpr: -7.2686***
  - fitb3m: -8.8063***

- PP test (With constant):
  - lwoil: -7.7661***
  - lwmaize: -7.7243***
  - lusgdp: -7.7909***
  - ffr: -5.9673***
  - lcpi: -8.6157***
  - lgdp: -27.8498***
  - lnneer: -9.7291***
  - ler: -10.6623***
  - cbkpr: -7.0882***
  - fitb3m: -8.1796***
- PP test (With constant and trend):
  - lwoil: -7.8293***
  - lwmaize: -7.7848***
  - lusgdp: -7.8535***
  - ffr: -6.1385***
  - lcpi: -8.7026***
  - lgdp: -27.3181***
  - lnneer: -9.5692***
  - ler: -10.5895***
  - cbkpr: -7.0380***
  - fitb3m: -9.5289***
- PP test (Without constant and trend):
  - lwoil: -12.6051***
  - lwmaize: -12.8448***
  - lusgdp: -10.6925***
  - ffr: -5.9924***
  - lcpi: -4.2727***
  - lgdp: -10.6758***
  - lnneer: -9.0580***
  - ler: -10.2001***
  - cbkpr: -7.1276***
  - fitb3m: -8.0649***

Notes on first-difference results:
- Almost all series show strong rejection of the unit root null at first difference (many entries significant at the 1 percent level, denoted by ***).
- Exceptions with lower significance include lcpi and lgdp in some specifications (e.g., ADF without constant and trend: lcpi -2.2617**; lgdp -2.1023**).

*Annex I. Table 1. Kenya: Results of Unit Root Testing — table notes report that statistical significance at 1, 5, or 10 percent levels (denoted by ***, **, *) would not allow accepting the null hypothesis.*

---

### Annex I. Table 2. Kenya: Estimates from Selected Specifications of the Single Equation Model

### Key estimated coefficients across specifications (columns (1)–(5))
- Constant:
  - (1): 0.02***
  - (2): 0.02***
  - (3): 0.02***
  - (4): 0.02***
  - (5): 0.01***
- Inflation (first lag):
  - (1): 0.79***
  - (2): 0.74***
  - (3): 0.79***
  - (4): 0.82***
  - (5): 0.75***
- Inflation (second lag):
  - (1): –0.15**
  - (2): –0.15*
  - (3): –0.18**
- Global oil price inflation (first lag):
  - (1): 0.01**
  - (2): 0.01**
  - (3): 0.02**
  - (4): 0.02***
  - (5): 0.17**
- Global maize price inflation:
  - (2) (first lag): 0.01*
  - (4) (first lag): 0.02**
  - (3) (second lag): 0.01**
  - (5) (second lag): 0.02*
- Change in NEER (first lag):
  - (2): –0.12*** (reported as a coefficient in one specification)
  - (4): –0.11*** (reported as a coefficient in one specification)
- Change in NEER — interacted with depreciation/appreciation dummy:
  - Interacted with dummy=1 for depreciation: –0.12**
  - Interacted with dummy=0 for appreciation: –0.10*
- Change in bilateral exchange rate (first lag):
  - (3): –0.08***
  - (4): –0.06**
- Change in short-term interest rate:
  - (3) (first lag): –0.001^
  - (4) (second lag): –0.001*

### Model fit and diagnostic statistics by specification
- Number of observations:
  - (1): 108
  - (2): 109
  - (3): 108
  - (4): 108
  - (5): 108
- Adjusted R-squared:
  - (1): 0.691
  - (2): 0.695
  - (3): 0.688
  - (4): 0.674
  - (5): 0.665
- F-statistics for joint significance of coefficients:
  - (1): 48.77***
  - (2): 47.02***
  - (3): 40.27***
  - (4): 45.17***
  - (5): 43.53***
- Jarque-Bera normality test:
  - (1): 4.94**
  - (2): 3.91
  - (3): 4.56^
  - (4): 10.91***
  - (5): 7.90**
- Durbin-Watson statistics:
  - (1): 1.93
  - (2): 1.76
  - (3): 1.93
  - (4): 1.94
  - (5): 1.74

Notes and interpretation pointers from the table:
- A positive change in exchange rate is an appreciation.
- Statistical significance at 1, 5, or 10 percent level (denoted by ***, **, *) would not allow accepting the null hypothesis that a specific parameter is zero.
- ^ denotes significance between 10 and 11 percent levels.

*Annex I. Table 2. Kenya: Estimates from Selected Specifications of the Single Equation Model — table notes as presented in the source.*

*Source: Annex I. Table 1 and Annex I. Table 2 from the provided IMF content unit.*

---


_Source: https://www.imf.org/-/media/files/publications/cr/2024/english/1kenea2024002.pdf_