The World Bank has yet to pin down a working definition of what it calls “private sector development (PSD).” In a recent paper, I argued that this helps explain why it repeatedly reorganizes its PSD work. New approaches to PSD appear, management concludes the existing organization no longer fits the latest approach, and another restructuring follows. Moving the boxes attempts to solve a definitional problem with an organizational solution.
That lack of direction in PSD is becoming more costly. Donor governments are cutting budgets sharply while 1.2 billion young people in developing countries will enter the workforce between 2025 and 2035. Private sector firms will have to shoulder most of the responsibility for creating the jobs these young people expect. Getting private sector development right means using public support to incentivize firms to invest and hire without entering markets where private investors would finance on their own and without subsidies set so high as to be wasteful.
In my earlier paper, I proposed a welfare test to anchor the World Bank Group’s PSD work. In a new paper, I examine what happens when the test is applied operation by operation.
Ratings of the PSD portfolio by the World Bank’s Independent Evaluation Group (IEG) barely moved as the bank shifted from directed credit to adjustment and privatization, the investment climate, jobs and inclusion, and the Cascade approach, which prioritized the mobilization of commercial private capital for development projects. Average PSD ratings varied by only 0.11 points across all five periods. The ratings showed no deterioration before the reorganizations and no improvement after them. The theories changed. The organizational boxes moved. The ratings stayed the same.
A test for private sector development
Can a definition of private sector development be derived from examining what PSD operations do? To answer this question, I start by collecting all operations the World Bank has classified under the Private Sector and Financial Development theme. This portfolio includes PSD operations that do quite different things with quite different instruments. No pattern emerges. IEG encountered much the same problem in 2016 when it reviewed the World Bank Group's private sector work and opted for an inclusive approach, lumping together innovation and entrepreneurship, business-regulation reform, housing finance, capital-market development, and public-private partnerships, to list a few, under the PSD umbrella. Between the PSD portfolio I compiled using the theme code and the PSD portfolio IEG constructed in 2016, there is no common operational definition holding these two portfolios together.
PSD is easier to abbreviate than to define.
In my paper, I define a private sector development operation as one in which public capital enters a market served by private suppliers to correct a market failure. When public money enters a market, it is likely to distort that market. So, a PSD operation must explain why an intervention is justified and what public financing does that the market would not do otherwise.
To assess whether a PSD operation should have received public financing in the first place, a separate test is applied. The test consists of three hurdles. An operation must clear each hurdle to merit public funding:
1. Justification: the operation appraisal document must identify a market failure that warrants the intervention.
2. Effectiveness: the instrument needs to fix the market failure. An instrument cannot clear this hurdle if it merely compensates for the market failure without actually fixing it. Similarly, the effectiveness hurdle is not cleared if an instrument sidesteps the market failure without fixing it.
3. Additionality: the appraisal must demonstrate that private capital would not, but for the bank’s intervention, finance the activity.
I apply this test to the existing portfolio of 156 operations coded as private sector development in FY24 and FY25. I excluded 45 of these operations because they had no appraisal document I could code. Of the remaining 111 operations, I excluded another 82 because they did not involve World Bank capital entering a market with existing private suppliers, or because they fell under a different additionality standard. This leaves 29 operations where World Bank capital could possibly compete with private sector suppliers.
Twenty-one of the 29 fail to clear at least one of the hurdles, and the remaining eight clear them all. Sixteen do not establish the market failure needed to justify intervention. Four use an instrument that does not change behavior at the point where the claimed market failure occurs. One does not establish financial additionality. The 21 failing operations account for $3.2 billion in commitments. Of this amount, $1.17 billion enters markets where suppliers are present, risking elbowing out private suppliers.
How much public support?
Once the case for public support is established, a second question arises: how much support is needed? An operation requiring a $10 million subsidy does not become better because the bank then agrees to provide a $30 million subsidy. The extra $20 million does improve one outcome: the recipient’s bank account is better padded. Competitive allocation approximates the minimum support the operation needs because private sector actors bid against each other to provide the operation with the smallest subsidy. In most situations, bilateral negotiations are unlikely to get private suppliers to reveal the minimum subsidy they require; asking firms how much subsidy they need is a wildly optimistic approach to price discovery.
I then tested whether independent readers applying the same protocol would select the same operations. This test covers which operations enter the sample, not the results of the three-hurdle test. I presented ChatGPT, Claude, and Gemini with the identical written protocol and the same set of 30 appraisal documents I used. This resulted in four independent readings of the data: one per language model plus my own. Each reader worked independently, without consulting the others. The results: all four agreed on nine of the 30 operations. Of these nine, three were included in the set of PSD operations and six were excluded. For the remaining 21 of the 30 operations, at least one reader disagreed with the others.
Disagreements among the four readers clustered in places where the original protocol left room for interpretation. Two issues accounted for the disagreements: how to treat budget support, and what constitutes a "substantially" private sector operation.
The World Bank has revised its approach to PSD many times and reorganized units responsible for it without ever settling on a stable operational definition, or a common rule to guide intervention in commercially supplied markets. Of the 29 operations included in the FY24–25 review where World Bank capital enters a market that is commercially supplied, 21 fail to clear at least one of the three hurdles on the strength of the evidence in their appraisals. Levels of subsidy are typically negotiated without reference to a rule about how much public support is needed.
Additionality should be a condition of each operation prior to approval, not an aspiration tagged onto PSD strategy. An Operational Policy for operations where World Bank capital enters a commercially supplied market should require the appraisal to identify the market failure and explain how the proposed instrument will change behavior where that failure exists. Additionality should occasionally subtract an operation from the portfolio. The appraisal should also demonstrate what the market would provide if the World Bank did not step in. Where the operation includes a subsidy, the World Bank should determine the minimum support necessary and set it using competitive pricing when possible. The proposed operational rules could stop the cycle of endless movement of boxes that try to solve definitional problems with organizational solutions.
