There Is No Dollar Shortage, Only a Wrong Price: Forex Rationing and the Queue Economy

The thesis stated plainly

Across the entire EPRDF period, and into the early years of Abiy Ahmed’s government, the central organising fact of Ethiopian economic life was the chronic shortage of foreign exchange. Importers waited months for letters of credit. Manufacturers reduced production for want of imported inputs. Hospitals ran short of imported medicines. Construction projects stalled. The Ethiopian press of the 2010s is full of articles about “the dollar shortage” — its causes, its consequences, its possible solutions. The Ministry of Finance, the National Bank of Ethiopia, and successive prime ministers all treated the shortage as the country’s central economic problem, requiring administrative solutions: foreign-currency allocation committees, sectoral priority lists, surrender requirements on exporters, import restrictions, and periodic crackdowns on parallel-market activity.

This article argues that there was no dollar shortage. There was, instead, a price at which the demand for dollars exceeded the supply, sustained by administrative means rather than allowed to clear. The technical name for what Ethiopia had is excess demand at an administered price. The colloquial name for it is queue economy. The shortage was an artefact of the policy, not a fact of nature, and it would have disappeared the moment the price was allowed to clear — as, in fact, it largely did when the 2024 float happened.

The argument has been made many times by Ethiopian and international economists. It has been made repeatedly in IMF Article IV consultations across the GTP decade. It is not original to this series. What this article does is set out the argument plainly, document the cost the policy imposed on the Ethiopian economy across thirty years, and ask the harder political-economy question of why the policy was maintained for so long despite consistent expert advice to abandon it.

The answer to that political-economy question is not that policymakers were stupid. The answer is that the queue economy served specific political functions for specific constituencies, and that the constituencies who benefited from it had more political voice than the ones who paid for it. The article makes this case explicitly.

How forex rationing worked

The mechanics of Ethiopia’s forex regime through 1992–2024 are worth describing because they show the disease in operation.

When an importer needed dollars to pay for goods purchased abroad, the importer applied to his bank for a letter of credit (LC) — a banking instrument by which the bank commits to paying the foreign supplier on the importer’s behalf, in foreign currency, in exchange for the importer’s commitment to repay the bank in birr. The application went through a sequence of approvals. The bank assessed the importer’s credit. The NBE determined whether foreign currency was available at the official rate. If available, the LC was issued; if not, the application went into a queue, with priorities determined by sectoral rules issued by the NBE.

The priority rules favoured strategic imports — fuel, fertilizer, pharmaceuticals, capital equipment for prioritised industries — over non-strategic imports (consumer goods, luxury items, certain raw materials, retail merchandise). The strategic categories shifted over time as policy priorities shifted. In any given year, a substantial portion of import applications either could not be funded at all or faced delays measured in months rather than days. By the 2020–24 period, documented waits for non-priority LCs ran from six to eighteen months, with some applications never being funded at all.¹

What this meant in practice is that the price of foreign currency in Ethiopia was not, as the official statistics claimed, the NBE-set rate. The effective price was the NBE-set rate plus the cost of the wait plus the cost of the paperwork plus the political-economy cost of maintaining the relationships needed to get into the priority queue. For an importer whose business could not survive a six-month wait, the effective alternative was either to shut down operations or to source the dollars through the parallel market at substantially higher prices. Many did the second. Some shut down.

The economic cost of this system was substantial in ways that the official statistics did not capture. Manufacturing-sector capacity utilisation in Ethiopia ran consistently below 50 percent through much of the 2010s, with input-supply problems frequently cited as the binding constraint.² Import compression — the suppression of import volumes below what the economy would have absorbed at market prices — was a continuous drag on growth. The Ethiopian Investment Commission’s own surveys of foreign-invested firms repeatedly identified forex availability as the single largest obstacle to operations.³

The Friedman-Sowell argument

The economic argument against the system was simple, repeatedly made, and consistently ignored by Ethiopian policymakers. A government that sets the price of foreign currency below the market-clearing level creates excess demand at the administered price. The excess demand has to be allocated by something other than price — by administrative rationing, by political connection, by queue position, by parallel-market access. Each of these allocation mechanisms is less efficient than the price mechanism would have been, because each conveys less information than a price about the actual value of the currency to different uses.

This is Hayek’s knowledge problem applied to foreign exchange. The market price of dollars would have transmitted, simultaneously and decentralisedly, information about the relative value of dollars for thousands of different uses: which importers would have paid most for them, which exporters would have parted with them most readily, which traders saw opportunities, which households needed to make remittance payments. The administrative allocation process could not transmit this information because it did not have it. The NBE’s forex committee had to guess at priorities and then enforce its guesses. The guesses were, inevitably, worse than the prices would have been.

This is Friedman’s monetary argument applied to the same problem. The exchange rate is, at root, a price — the price of one currency in terms of another. Like all prices, it conveys information about supply and demand. Suppressing the information does not change the underlying conditions; it only hides them. The parallel-market premium that grew under the queue economy was the information leaking out through the channels the state could not suppress. It was, as article 8 argued, the honest rate.

This is Sowell’s political-economy argument applied to the same problem. The queue economy was politically functional because it created rents for the entities that could access it. The importers who could get LCs at the official rate were getting dollars at a discount to the market price; the difference was a transfer to them. The state-owned enterprises that received priority allocations were getting subsidised inputs. The politically-connected private firms in the priority sectors were getting cheap capital equipment. The parallel-market dealers were getting margin on the difference between official and unofficial rates. Each of these constituencies gained from the queue. The losers — the small private importer, the manufacturer who could not get inputs, the consumer who paid more for goods produced at low capacity utilisation — were diffuse, unorganised, and politically inaudible.

The exporter side

The other side of the regime, less discussed but equally consequential, was the surrender requirement on exporters. Ethiopian exporters of coffee, sesame, oilseeds, leather, gold, and other commodities were required to sell a percentage of their foreign-currency earnings to the NBE at the official rate. The surrender percentage varied over time: 70 percent at some points, 50 percent at others, with the remainder available to the exporter for input purchases or for sale on a secondary market.

The surrender requirement was, formally, a tax on exports. The tax rate was the difference between the parallel-market rate and the official rate, applied to the surrendered share of the exporter’s earnings. At a parallel premium of 50 percent and a surrender requirement of 70 percent, the implicit export tax was 35 percent of foreign-currency earnings — a substantial tax that, predictably, suppressed export activity.

The behavioural responses were also predictable. Exporters under-invoiced their exports, retaining the difference in offshore accounts. Exporters routed transactions through neighbouring countries to avoid the surrender requirement. Some exports — gold in particular — were smuggled in volumes that, by some estimates, exceeded the official export figures.⁴ The Ethiopian state, by imposing a heavy tax on exports through the exchange-rate mechanism, succeeded in reducing the supply of foreign currency it was simultaneously trying to ration.

This is the deep diagnostic point. The “dollar shortage” was, in part, produced by the very mechanism that was supposed to manage it. Exporters who would have brought more dollars into the formal banking system at a market rate were instead routing dollars around the system because the surrender requirement made the formal channels uneconomic. The shortage and the surrender requirement were mutually reinforcing: the requirement reduced supply, which made the shortage worse, which justified maintaining the requirement to ration what supply remained.

The political constituencies

If the queue economy was so demonstrably costly, why was it maintained for thirty years against consistent IMF, World Bank, and academic advice? The answer is the political-economy core of this series, and it deserves direct statement.

The queue economy served three constituencies. State-owned enterprises received subsidised foreign currency through the priority-allocation mechanism; ending the queue would have meant they had to compete for dollars at market prices, raising their input costs and reducing their financial performance. Politically-connected private firms in priority sectors received the same subsidy; ending the queue would have made them pay market prices for dollars they had been getting at administrative prices. Parallel-market participants — the dealers, hawala networks, and informal-sector traders who arbitraged the gap between official and parallel rates — earned substantial rents from the gap; ending the queue would have eliminated the gap and the rent with it.

Each of these constituencies had political voice. The state-owned enterprises were closely tied to the ruling party. The politically-connected private firms were, in many cases, owned by individuals with personal relationships to ruling-party leadership. The parallel-market participants were less politically organised but had substantial economic significance and could exert pressure through their role in financing political activity (a relationship that has been documented in multiple Ethiopian political-economy studies though never formally proven).⁵

Against these three constituencies, the losers from the queue economy — the small private importer, the manufacturer who could not get inputs, the urban consumer who paid more for goods made at low capacity utilisation, the diaspora household whose remittance bought less than it would have at a market rate — were diffuse and politically unorganised. Each individual loss was small relative to the gain to the beneficiary. No political coalition formed to demand the queue’s end. The political economy of the queue was structurally biased toward its persistence.

This is a Sowellian point. It is not that anyone consciously decided to maintain a costly system. It is that, given the way costs and benefits were distributed across different political constituencies, the politically rational decision for any particular government was to maintain the system, manage its visible costs, and address the underlying problem only when external pressure (an IMF programme, a sovereign-debt crisis) forced the issue. The 2024 float happened, not because the political economy of the queue had changed, but because the macroeconomic situation had deteriorated to the point that the IMF programme — and the financing it brought with it — was a higher priority than maintaining the queue.

What the 2024 float showed

The post-July-2024 evidence on this question is, in the series’s reading, decisive. When the NBE moved to a market-determined exchange rate in late July 2024, the consequences for the “dollar shortage” were immediate and dramatic. Within ten days, the birr had moved from 57 to roughly 114 against the dollar — a 100 percent depreciation that closed most of the gap with the parallel market.⁶ By mid-2026, the rate had drifted further to approximately 157, with the parallel premium narrowed to approximately 12 percent.

What disappeared along with the gap was the shortage. Letters of credit, which had been queued for months, began being issued in days. The IMF noted in its first review of the ECF programme in October 2024 that “the exchange rate in the official market has largely closed the gap to the parallel market, with little disruption to the broader economy.”⁷ Import volumes, which had been suppressed for years, began to recover. The state-owned-enterprise sector adjusted — painfully, in some cases — to higher input costs. The Ethiopian Investment Commission’s own subsequent surveys showed that forex availability had ceased to be the top reported obstacle to foreign-invested firms’ operations.

The shortage did not disappear because dollars had become more plentiful. The actual stock of foreign-currency reserves rose only modestly in the immediate post-float period; the underlying supply of dollars to the economy was largely unchanged. The shortage disappeared because the price at which dollars were offered had risen to the point where demand and supply equilibrated. The shortage had been an artefact of the price. The float removed the artefact.

This is the empirical vindication of the article’s thesis. The dollar shortage was never a structural feature of the Ethiopian economy; it was a structural feature of the Ethiopian policy regime. Different regime, different shortage — or, in this case, no shortage.

The honest counter-case

The strongest version of the defence of the pre-2024 regime is that maintaining the official exchange rate served broader macroeconomic objectives: it kept inflation lower than it would otherwise have been (by suppressing the import-price pass-through), it gave the state policy space to direct credit and investment to priority sectors, and it provided a degree of stability against the volatility a free-floating exchange rate would have introduced. The float, on this account, was undertaken only when the macroeconomic situation had deteriorated so far that the costs of maintaining the regime exceeded the costs of abandoning it.

The series’s reply is that this defence underweights the cumulative cost of the regime across thirty years. The benefits of suppressed import-price pass-through were real but bounded; the costs in import compression, in capacity underutilisation, in suppressed exports, in the parallel-market premium, in the political economy of patronage that the regime sustained — these costs were larger than the inflation-suppression benefits, by margins that the IMF Article IV consultations and the post-2024 evidence have made fairly clear. The regime was maintained not because its benefits outweighed its costs in an honest cost-benefit accounting, but because the costs fell on politically unorganised constituencies while the benefits accrued to politically organised ones.

The next article turns to the parallel monetary system that the queue economy generated alongside itself — the diaspora’s shadow currency, the hawala networks, and the de facto dollarisation of the elite economy that filled the space the official system could not.

---

Footnotes

1. International Monetary Fund, “Ethiopia: 2024 Request for an Arrangement Under the Extended Credit Facility,” IMF Country Report 24/241 (July 2024), discussing pre-reform LC waiting times. ↩

2. World Bank, Ethiopia Country Economic Memorandum (World Bank, 2019); Central Statistical Authority of Ethiopia, Manufacturing Survey various issues 2010–2020. ↩

3. Ethiopian Investment Commission, Foreign Direct Investment in Ethiopia: Annual Surveys 2015–2020. ↩

4. African Center for Strategic Studies, “Conflict Gold: Ethiopia’s Smuggling Problem,” 2020; Brian Lawson, “Ethiopia’s Disappearing Gold,” Foreign Policy, May 2022. ↩

5. Sarah Vaughan and Mesfin Gebremichael, “Rethinking Business and Politics in Ethiopia: The Role of EFFORT, the Endowment Fund for the Rehabilitation of Tigray,” Africa Power and Politics Programme Research Report 2 (2011). ↩

6. PWC Kenya, “Ethiopian Birr Devaluation,” August 2024. ↩

7. International Monetary Fund, “Ethiopia: First Review under the Extended Credit Facility Arrangement,” IMF Country Report 24/302 (October 2024). ↩