There is nothing special about the number 4,000. Economics has no rule that says a shilling is in danger at 4,000 to the dollar but safe at 3,990.
But when Uganda’s currency crossed that line in October, many called it a ‘psychological barrier’, and few disagreed.
Eight days earlier, the Central Bank had told the public there was no need to panic. Then came an emergency meeting that raised the main interest rate from 9.75 percent to 12 percent.
That line captures one of Central Banking’s oldest puzzles.
A Central Bank sets one price, the interest rate. But what it wants to move is invisible, which is what millions of people believe will happen next.
Central Banks usually learn what people believe only after they have acted on it.
What is the problem?
People set prices according to what they expect. A shopkeeper who expects the shilling to keep falling raises prices today to cover tomorrow’s dearer imports.
A worker who expects inflation (a general rise in prices) asks for a bigger pay rise. A family fearing a weaker Shilling swaps its savings into dollars, weakening it further. Each time, a belief about the future changes behaviour now, and helps make itself come true.
This is why Central Banks talk so much about ‘expectations’. If people trust that prices will stay steady, a shock such as higher fuel prices passes and fades. If not, it can become a lasting spiral. The interest rate is only half the central bank’s toolkit. The other half is persuasion.
The trouble is that Central Banks cannot see what they are trying to persuade. What people believe today is hard to measure, and even their hard data arrives late.
Uganda’s inflation figures, for example, appear around each month’s end, but output figures for October to December come only at the end of March.
Worse, a rate decision takes ‘one to two years’ to reach prices, as the late governor Emmanuel Tumusiime-Mutebile put it in 2014, so the Central Bank must act on what people will do, not what they did.
How and why it bites
The problem has three layers, each worsening the next.
The first is inattention. Most people have jobs, families and bills, not Central Bank statements, on their minds.
Economists who surveyed firms in New Zealand, which pioneered inflation targeting, found that after nearly 25 years of it, most bosses paid little attention to inflation, and their estimates of it were scattered and often far too high. If that is true of managers in a rich country, it is truer of market traders in Kampala.
The second is the messenger. Most people hear about policy second-hand, through newspapers, radio and gossip.
An American experiment with nearly 20,000 households found that reading the Federal Reserve’s actual statement shifted people’s inflation expectations, but reading news reports of the same meeting had only about half the effect.
Each retelling loses some of the message.
The third is speed. In 2023, depositors at Silicon Valley Bank in California tried to withdraw $42bn in a single day after worries spread through group chats.
In Uganda, the value of mobile-money payments rose by 28.4 percent in the year to September 2025. The problem is that beliefs shift in hours and official data take months.
Uganda has a fourth reason the problem bites harder: the Central Bank’s main lever works mostly through markets and beliefs, not bank loans.
Two International Monetary Fund (IMF) economists, Hany Abdel-Latif and Geoffrey Bannister, found that when the Bank of Uganda raises its rate by one percentage point, the rates banks charge borrowers barely move at first and rise only about 0.7 points in the long run.
Lending hardly responds. Bank lending rates have hovered around 20 percent since 2011, whatever the Central Bank does.
Treasury bill rates, by contrast, move about 0.7 points almost at once, which matters because the state borrows heavily, so investors’ mood hits the budget directly.
According to the third Charter for Fiscal Responsibility for FY2026/27-FY2030/31, the percentage of government revenue in 2025/26 that went to interest rates is at 32.5 percent, against about 12 percent for the typical sub-Saharan country.
Earlier research found that the exchange rate is Uganda’s main channel.
Taken together, the evidence suggests policy works through investors, the shilling and expectations, so reading the mood matters more in Uganda than in countries where bank loans do the work.
Together, these layers create a dangerous gap. The public barely watches, hears the message distorted, and changes its mind faster than the Central Bank can notice. Crises are born there.
What Uganda has tried
In June, two Bank of Uganda analysts, Phillipa Mukasa, a financial data analyst, and Joseph Ssentamu Nyanzi, who heads domestic situation analysis and macroprudential data management, published a working paper that tries to measure the public mood directly.
Traditional indicators, they write, ‘may not fully capture rapid changes in confidence, risk perceptions, and market narratives that often emerge during periods of stress or optimism.’
They used FinBERT, a program that scores financial writing as positive, negative or neutral, on 4,286 economic articles from five Ugandan outlets (2019-2025) to build a ‘public sentiment’ index, and on 140 central bank and finance ministry publications for a ‘policy sentiment’ index.
The idea is to treat sentiment ‘as both a reflection of prevailing macro-financial conditions and a potential transmission channel through which expectations influence financial outcomes.’
The results were mixed but useful. Press mood tracked bank health: cheerful when banks were profitable, gloomy when bad loans rose.
But the authors admit the links are ‘economically small’, and press mood could not forecast quarterly bank figures. Its strength lay elsewhere.
They describe public sentiment as ‘an early, reactive signal of emerging stress’, one that turned ahead of big shifts such as the arrival of Covid-19.
The official index behaved differently. Its tone changed before interest rates did, which they take as support for the view that the Central Bank’s words work as ‘an active expectations-management tool.’
Sentiment, they caution, should be ‘interpreted as an informative signal rather than a literal measure of ‘true’ beliefs’, partly because FinBERT ‘is trained largely on advanced-economy financial text’ and may miss local turns of phrase.
Both indices, they propose, should sit on the Central Bank’s dashboards as ‘complementary ‘context layers”, not as alarms that trigger action on their own.
The IMF study shows how much the shilling already drives policy.
Between 2011 and 2023, it found a sudden 15 percent fall in the shilling typically led the Central Bank to raise its rate by up to half a percentage point within three months, and such currency shocks explained about 12 percent of the swings in the policy rate.
Some shocks came from news rather than economics, such as the World Bank’s response to the Anti-Homosexuality Act. Headlines move the Shilling, and the Shilling moves policy.
This time the reaction was far larger and came twice.
On September 24, the apex bank raised the share of deposits banks must park with it from 11 to 13.5 percent.
Three weeks later, it added 2.25 points to the key policy rate. With September inflation at 4.6 percent, still below target, the Central Bank was ahead of prices but was behind the market because the shilling had already crossed 4,000.
The emergency policy statement, read by the governor, Michael Atingi-Ego, was mostly about belief.
It warned that rising prices could take hold ‘if monetary policy does not contain these pressures now,’ and cited the Central Bank’s own surveys showing that consumer sentiment had ‘weakened marginally’.
His aim, he said, was to stop people expecting rising prices and a falling shilling, beliefs that are hard to shift once settled.
He said the meeting grew out of concerns that ‘the Central Bank had and the public.’
The Central Bank acted partly on public mood, but had no daily way of measuring it.
What the rest of the world does
The oldest tool is the survey.
The European Central Bank’s monthly Consumer Expectations Survey, started in 2020, asks about 19,000 people in 11 countries what they expect for prices, jobs and their finances.
The Central Bank of Kenya runs a Market Perceptions Survey of banks and businesses before every rate-setting meeting.
Surveys are reliable but slow and costly, and capture what people say, not what they do.
The newer tool is text. The Federal Reserve Bank of San Francisco publishes a daily index of the mood in the news.
Spain’s Central Bank found its newspaper index tracked the economy better than official confidence surveys.
Italy’s built daily measures of inflation expectations from Twitter posts that beat other sources at predicting its household surveys.
Listening is only half the job.
When the Bank of England tested its messages on more than 2,000 people, a summary in everyday language, with relatable examples such as what the 2 percent inflation target means for shop prices, raised understanding by 40 percent over its usual technical statement.
The American experiment agreed, arguing that simply telling people the Fed’s inflation target worked as well as the full statement.
The leaders share habits. They measure mood often and from several sources, test whether it predicts anything, write for the public, and treat it all as an aid to judgement.
Is the problem solved?
No. It has been made smaller, not solved. The most thorough recent review, by Alan Blinder, a former vice-chairman of the Federal Reserve, and colleagues in the Journal of Economic Literature, concluded: ‘We see some promise in the potential to inform the public better, but many challenges along the way’.
Its title, tellingly, is ‘Promise or False Hope?’
Three gaps remain everywhere.
The first is measurement. Surveys are slow, and Georgia Tech researchers have shown that mood-reading software struggles with central-bank language, where the real question is whether rates will rise or fall.
The second is interpretation. Before the 2008 crash, the Bank of England found the warning sign was not gloom but too much cheer.
The third is reach. Most people still hear the central bank through the media.
In Uganda, the gaps are wider. There are few household surveys and shallow markets, and much economic talk happens in local languages, on radio and WhatsApp, where today’s tools barely listen.
As Ms Mukasa and Mr Nyanzi note, the models were built for a different world.