Financial Repression as a Hidden Tax: The Developmental-State Years

The tax that did not appear on the budget

Every state taxes its citizens. Most do it openly: income taxes, value-added taxes, customs duties, property taxes, all listed in budgets that get debated, often loudly, in parliaments and newspapers. The Ethiopian state did this too, of course. But across the entire EPRDF period — from 1991 to 2018, and in modified form well beyond — the largest single transfer of resources from the population to the state-favoured economic sectors did not appear in any budget. It did not appear because it was not, formally, a tax. It was the difference between the interest rate Ethiopian savers earned on their deposits and the rate of inflation, multiplied across the entire deposit base of the banking system, year after year. It was the implicit tax on savers that financed the credit subsidies received by state-owned enterprises, infrastructure projects, and politically connected private borrowers.

The technical name for this mechanism is financial repression. The standard reference is the Shaw-McKinnon framework developed in 1973, which described how developing-country governments could mobilise resources for state priorities by holding deposit rates below market-clearing levels.¹ This article argues that Ethiopia under the EPRDF ran one of the most sustained and consequential financial-repression regimes in modern African history, and that the cost — borne disproportionately by Ethiopian savers who had no political voice in the arrangement — was a major contributor to the chronic monetary disorder this series is tracking.

The argument has a moral dimension that is worth stating openly. A tax that funds public goods and that is openly debated in parliament can be defended on democratic-legitimacy grounds even when the citizens taxed dislike it. A tax that no one acknowledges is a tax, that funds projects the taxpayers had no say in selecting, that falls disproportionately on those least able to defend themselves — this is harder to defend. The financial-repression machinery of the EPRDF years was, in this sense, a regressive transfer in the dark. The series’s editorial position on it is critical, and this article will defend that position.

The mechanism

How financial repression operates in technical terms is worth setting out clearly, because the mechanism is invisible to most participants in the system.

Step one: the central bank sets a ceiling on the interest rates that commercial banks can pay on deposits. In Ethiopia under the EPRDF, the NBE-set minimum deposit rate fluctuated between 3 and 7 percent for most of the period.² This is the rate Ethiopian households and small businesses earned on their savings.

Step two: inflation runs above this rate. Ethiopian official inflation averaged 8 to 15 percent through most of the EPRDF period, with extended episodes above 20 percent (2008–11) and acute episodes above 30 percent (2022–24).³ The parallel-rate-implied “true” inflation was often higher.

Step three: the gap between the deposit rate and the inflation rate is the real interest rate on savings. For Ethiopian savers, this real rate was negative for most of the EPRDF period — often substantially so. A saver who deposited 1,000 birr in 2010 at the prevailing 5 percent rate would, by 2015, have a deposit worth 1,276 birr in nominal terms; the same goods that 1,000 birr would have bought in 2010 would, by 2015, cost roughly 1,650 birr.⁴ The real purchasing power of the deposit had fallen by roughly 23 percent over five years. The saver had been quietly taxed.

Step four: the banking system, having mobilised these deposits at sub-inflation rates, lends them on at administratively-determined rates that are also below market-clearing levels — but to a different population. The borrowers were, principally, state-owned enterprises (the EEPCO, Ethio Telecom, Ethiopian Airlines, the Industrial Development Holding Company and its subsidiaries), state-favoured infrastructure projects (the Addis–Djibouti railway, the Grand Ethiopian Renaissance Dam, the industrial parks), and politically connected private borrowers in priority sectors (sugar, cement, textiles).⁵ These borrowers received credit at rates that, after adjusting for inflation, were often also negative — meaning the state-favoured borrower could borrow money and make a profit on the loan simply by holding the proceeds while the real value of the debt eroded.

Step five — and this is where the architecture’s redistribution becomes visible — the difference between the negative real rate paid to depositors and the (also negative, but smaller in absolute value) real rate charged to state-favoured borrowers is the implicit subsidy that the banking system delivers to the state priorities. The subsidy is funded by the depositors. The depositors are not consulted. The recipients are chosen by the state. The mechanism is a tax-and-spend operation that operates entirely outside the formal budget.

The 27 percent rule

The most visible Ethiopian instrument of financial repression — the one that drew the most attention from foreign analysts and international financial institutions — was the so-called “27 percent rule.” Introduced by NBE Directive in 2011 and refined in subsequent directives, the rule required private commercial banks to invest 27 percent of every loan they extended into NBE-issued bonds (the “DBE bonds” or “NBE bills”) at administratively-set interest rates below the rates the same banks would earn on commercial lending.⁶

The mechanism is worth understanding because it is the mirror image of the standard reserve-requirement tool. A normal reserve requirement says: hold a fraction of your deposits in non-interest-bearing form at the central bank. The Ethiopian 27 percent rule said: hold a fraction of your loans in low-interest-bearing form, in bonds issued by a state development bank, whose proceeds the state bank could then lend to state priorities.

In effect, the rule forced private banks to fund state development priorities — to lend 27 cents of every dollar they nominally extended to private borrowers to the state, via the DBE, instead. The interest rate on the bonds (initially 3 percent, gradually rising) was substantially below the rate the private banks would have earned on the private lending they were forgoing. The gap was the implicit tax on private banks, which they passed through to their customers in the form of higher lending rates and lower deposit rates.⁷

The 27 percent rule was finally relaxed and then eliminated in 2019 under Abiy Ahmed’s early reform programme. The IMF reviews of the Ethiopian banking sector through the 2010s consistently flagged it as a major source of misallocation of credit and a tax on private financial development.⁸

The directed-lending machinery

The rule was, however, only one piece of a larger machinery. The Commercial Bank of Ethiopia, dominant in the deposit base throughout the period, was the principal vehicle for directed lending to state priorities. CBE’s loan book at various points in the 2010s included tens of billions of birr in loans to state-owned enterprises — many of them non-performing in any normal accounting sense but never recognised as such on the bank’s books.⁹

The Development Bank of Ethiopia served a parallel function for medium- and long-term industrial lending. DBE’s mandate was explicit: provide credit to manufacturing, commercial agriculture, and other priority sectors at rates designed to support investment rather than to clear markets. The DBE was funded principally by NBE refinancing — that is, by the central bank — and by the 27-percent-rule proceeds from private banks. The DBE’s loan book through the 2010s was, by the bank’s own reporting, heavily concentrated in a small number of large projects, many of them ultimately unsuccessful.¹⁰ The non-performing-loan ratio at DBE rose sharply through the 2010s and necessitated a substantial recapitalisation in 2022–24 as part of the post-float reform programme.¹¹

The Industrial Development Holding Company, established in 2014 and renamed Ethiopian Investment Holdings in 2022, consolidated state-owned commercial assets into a sovereign-wealth-style holding structure that, in effect, sat between the banking system and the state-owned enterprises. The structure obscured the consolidated picture of state-enterprise indebtedness, but its accumulated losses required substantial state-bank financing throughout the period.¹²

The distributional consequences

Who pays the financial-repression tax, and who benefits from the directed lending?

The taxpayers are, principally, holders of birr-denominated savings. This group includes urban salaried workers with deposit accounts, pensioners drawing from accumulated savings, small businesses holding working capital, and any household sufficiently formal-sector-engaged to use the banking system. The poor, who hold their savings in cash, livestock, or equb informal-credit-society contributions, are partially protected from the tax — they get the inflation cost but not the negative real interest rate cost. The middle class and upper-middle class, with deposit accounts and bond holdings, are the principal taxpayers.

The beneficiaries are the recipients of the cheap credit: state-owned enterprises, infrastructure-project contractors, and the politically-connected private borrowers who can access the formal banking system at administrative rates. Among private borrowers, this group is overwhelmingly composed of large firms — those with the political connections to qualify for priority-sector designation and the size to access the rationed credit. Small and medium private firms, lacking these connections, were largely excluded from the formal banking system throughout the EPRDF period.

The net effect is a transfer from middle-class urban savers to large state and politically-connected private firms. This is, by any reasonable definition, a regressive transfer — moving resources from a broader, less-wealthy base to a narrower, wealthier one. The defence offered by the developmental-state framework is that the transfer funds productive investment that generates broad-based growth which, over time, benefits everyone. The empirical question is whether the productive investment has, in fact, generated the predicted broad benefits. Article 14 examines that question in detail. The short answer here is: partially.

The double-counting of capital

There is a deeper economic distortion the financial-repression regime generated, beyond the distributional one. When real interest rates are negative — when borrowers are paying back less in real terms than they borrowed — the cost-benefit calculus on which capital investment is made gets distorted in a specific way. Projects that would not be viable at market-clearing real interest rates become “viable” at negative real rates, simply because the implicit subsidy from the financial system makes them appear so. The Addis–Djibouti railway, the GERD, the industrial parks, the sugar projects, the large-scale commercial farms — many of these were undertaken on the assumption of cheap, available credit.

When the credit subsidy is partial or temporary, the investment may still pay off. When the subsidy is sustained over a decade and a half, however, the result is a substantial accumulation of investment in projects whose underlying economic returns are insufficient to service their debt at market rates. The result is what Hayek and other Austrian-tradition economists call malinvestment — capital allocated to uses that do not generate the returns required to justify them, sustained only as long as the subsidy continues.¹³

When the subsidy ends — as it began to end in the post-2018 reforms and ended decisively with the 2024 float — the malinvestment becomes visible. The state-owned enterprises whose debt was sustainable at 3 percent real rates become insolvent at 12 percent real rates. The infrastructure projects whose user fees were calibrated to a 57 birr/USD exchange rate become unable to service dollar-denominated debt at 157 birr/USD. The 2023 Eurobond default and the 2022–24 CBE recapitalisation are, in part, the bills for the financial-repression years.

The honest counter-argument

A serious case must engage the strongest version of the financial-repression defence. The defence runs as follows: developing countries with shallow capital markets, low domestic savings, and limited tax capacity need mechanisms to mobilise resources for long-horizon investment. The market, left alone, will not provide these resources at the scale required. Financial repression — used by Japan in the 1950s and 1960s, by South Korea in the 1960s and 1970s, by China across the 1980s and 1990s — has been one of the principal mechanisms that has, in fact, worked to deliver developmental investment in real cases.¹⁴ To single Ethiopia out for criticism on this front is to apply a standard that the most successful developmental states would also fail.

This is a serious objection and the series takes it seriously. The reply has two components.

First, the East Asian financial-repression regimes had features that the Ethiopian version did not. They were time-limited — relaxed as the economies industrialised and capital markets deepened. They were export-disciplined — the credit subsidies went to firms required to compete in export markets, which provided an external test of efficiency that the Ethiopian system did not impose on its beneficiaries. They were politically accountable in specific ways — the South Korean system, for instance, terminated credit subsidies to firms that missed export targets, in a way the Ethiopian system did not. The mere fact that East Asia used financial repression does not establish that Ethiopia’s specific implementation of financial repression was equivalently effective. The features that made East Asia’s system work are largely absent from the Ethiopian one.

Second, even East Asia’s successes came at substantial cost — the South Korean chaebol system that survived the financial-repression years has been a chronic source of political-economy distortion ever since, the Japanese banking system’s accumulated bad loans took two decades to resolve, the Chinese state-owned-enterprise sector is, in 2026, a continuing macroeconomic problem of substantial magnitude. The point is not that financial repression cannot be a useful developmental tool. It is that it has identifiable costs, and that those costs need to be weighed against the benefits in any honest accounting.

For Ethiopia, this article argues, the accounting comes out unfavourably. The benefits are real but partial; the costs — in the form of accumulated bad loans, in the form of crowded-out private credit, in the form of accumulated dollar-denominated debt now needing to be serviced at post-float exchange rates — are large and durable. The financial-repression machinery did some real work. It also imposed a hidden tax on a generation of Ethiopian savers, and the bill is still being paid.

The next article turns to the institutional feature that, in some ways, mattered more than any of this: the absence of a real credit market for the majority of Ethiopians who lived on land they did not own.

---

Footnotes

1. Edward S. Shaw, Financial Deepening in Economic Development (Oxford University Press, 1973); Ronald I. McKinnon, Money and Capital in Economic Development (Brookings, 1973). ↩

2. National Bank of Ethiopia, Quarterly Bulletins, various issues 2000–2018, summarised in IMF Article IV consultations, particularly IMF Country Report 18/354 (December 2018). ↩

3. World Bank, “Inflation, consumer prices (annual %) — Ethiopia,” World Development Indicators; Trading Economics, Ethiopia inflation historical series. ↩

4. Calculation based on cumulative inflation 2010–2015 from World Bank WDI Ethiopia series and NBE deposit-rate minimums. ↩

5. For the composition of CBE and DBE loan books over the period, see World Bank, Ethiopia Financial Sector Development Report (World Bank, 2019); IMF Country Report 18/354. ↩

6. National Bank of Ethiopia, Directive SBB/55/2013 and amendments. Discussed in Tony Addison, Olu Ajakaiye, and Tilman Brück, “The 27 Percent Rule: Financial Repression in Ethiopia,” WIDER Working Paper, 2016. ↩

7. IMF, “Ethiopia: 2014 Article IV Consultation — Staff Report,” IMF Country Report 14/303 (October 2014). ↩

8. IMF Country Reports 14/303, 16/322, and 18/354. ↩

9. World Bank, Ethiopia Financial Sector Development Report (2019). ↩

10. Development Bank of Ethiopia, Annual Report 2016/17, 2017/18, and 2018/19 issues. ↩

11. International Monetary Fund, “Ethiopia: 2025 Article IV Consultation,” July 2025, on the CBE and DBE recapitalisation requirements. ↩

12. Ethiopian Investment Holdings, founding statute, https://eih-eth.com; IMF Country Report 25/188 (July 2025) on EIH transparency benchmarks. ↩

13. For the Austrian theory of malinvestment, see Friedrich A. Hayek, Prices and Production (Routledge, 1931); Roger W. Garrison, Time and Money: The Macroeconomics of Capital Structure (Routledge, 2001). ↩

14. For the East Asian financial-repression record, see Joe Studwell, How Asia Works: Success and Failure in the World’s Most Dynamic Region (Profile, 2013), particularly chapter 3. ↩