If You Want Clients to Borrow from the World Bank for Net Zero Investments, Make It Easier

High-income countries trying to reach net-zero targets are confronting the fact that existing environmental regulations make it harder to build low-carbon infrastructure. Lobbying groups in the US, for example, are using regulatory tools to block geothermal, solar, and wind facilities. Providing subsidies and incentives for low carbon isn’t enough: revisiting, rewriting, and in some cases retiring these regulations is going to be a vital part of faster progress toward the green economy.

The issue isn’t unique to high-income countries, and it is perhaps particularly important when it comes to climate finance. That’s because multilateral development banks including the World Bank are being asked to do more on climate, but they are constrained by their own bureaucracy, including a set of environmental regulations that add costs and delays to investments that could support global mitigation goals.

Look at answers to World Bank client surveys in some of the countries where you might hope the institution would have a big role to play in financing mitigation projects.  When asked what the World Bank’s greatest weakness is, the top answer in Brazil: “processes too slow and complex.” That’s the same top answer as Indonesia and Vietnam. Complex and slow processes rank second out of sixteen and in India and Argentina, and fourth in South Africa and Türkiye. Good project preparation takes time, and rushed projects tend to have worse development outcomes. Nonetheless, there clearly can be too much of a good thing in terms of bureaucracy.

The bureaucratic hassles of borrowing from the World Bank are surely one factor behind seemingly low demand from clients for IBRD lending outside of times of crisis.  (Dated) survey evidence suggests that client countries view the Bank as a lender of last resort with infrastructure in particular because of the costs of doing business, something also reported by World Bank staff in a 2010 Independent Evaluation Group survey. The response shouldn’t be to take scarce grant resources and use them to fund inefficient subsidies to drum up demand for climate lending in IBRD countries, it should be to reform lending processes and approaches to make borrowing from the IBRD for climate mitigation at the institution’s unsubsidized (but below-market) rates more attractive.

Part of that process should involve looking at lending instruments and safeguards. The World Bank has two main financing instruments: broad policy lending and specific investment financing.  It also divides potential projects into four main categories based on potential environmental impacts: Category A; where the potential impact is large and borrowers and the bank need to undertake assessments and agree mitigation measures; Category B which is lower risk but still requires assessment and mitigation; and Category C which is likely to have minimal or no adverse environmental impacts. A separate category, F, involves projects where funds flow through a financial intermediary.

We can look at the impact of project type and environmental assessments on the bureaucratic burden of project preparation by comparing the length of time it takes to get a project from initial idea to World Bank board approval. Analysis by Christopher Kilby suggests that project preparation times are primarily a function of World Bank processes rather than recipient country characteristics, down to factors including loan type and amount (and US interest in the client country). Kilby’s results suggest Structural Adjustment Programs and Development Policy Loans take 240 days shorter to prepare than investment projects, while a project in a country directly represented by an Executive Director on the World Bank’s board is associated with a 52 day reduction in preparation time. In a later paper with Kevin Gallagher, Kilby looks at the impact of Environmental safeguard categorization and suggests projects have a typical preparation time of 293 days with no safeguards and 416 days with safeguards, though there is some evidence of convergence in preparation times with more recent projects.

We use the database examined by Kilby to look at the impact of environmental assessments in particular, updated through 2021. One missing element in that database (as reported by Kilby) is the date when the project was conceived, but luckily World Bank project ID numbers, which are in the database, are issued sequentially. That means we know project 134156 was greenlighted for preparation by World Bank management just after project 134155 and just before project 134157, for example.  We exclude projects prior to 1994 and project IDs below 20,000 (which do not follow a fully sequential numbering system) as well as recent projects (project number above 175,000), because the slower projects won’t have reached the Board yet. We drop earlier loans without an approval date (themselves dropped from the project pipeline). And we drop loans without an IBRD or IDA commitment amount, in order to focus on traditional World Bank projects. We then run the following regression:

[Approval date] - [average approval date of 100 closest below and above projects by ID]


[Total IBRD + IDA commitment] + [(Exclusive) IDA financing dummy] + [IDA/IBRD blend dummy] + [Instrument (structural adjustment/policy lending dummy)] + [Sector/theme of infrastructure] + [Environmental assessment category dummy (one each for A, B, F, and other)]. 

The sector/theme dummy is a 1 if the project theme/sector codes mention any of the words energy, power, water, road/roads, transport, irrigation, port/ports, airport/airports, infrastructure, construction, or housing, and 0 otherwise.

The regression results report how many days, more or less, a project with particular features takes to make it to the World Bank board for approval than the average of projects initially conceived at around the same time. The first column reports results for the whole sample, the second column for the second half of the sample (with the earliest Board approval in 2008), and the third column for the last quarter of the sample (with earliest board approval in 2015).

The results suggest that:

  • Project size (commitment amount) is not a major factor in length of project preparation (a $100 million project will take about nine more days than a $10 million project).
  • Compared to projects initiated at about the same time, those projects that are purely finance by IDA as compared to IBRD projects take 71 fewer days to reach Board approval, although that effect is not apparent more recently (when IDA/IBRD blend projects appear to take longer).  
  • Policy lending takes 107 fewer days to reach the board than investment lending (72 days less in the most recent sample).
  • Projects that involve infrastructure do not take longer to be approved—this allowing for other features included in the regression.
  • Environment category F and B add about 100 days to project preparation while category A projects take nearly a year longer to reach the board, although delays have declined over time (to 43 days in the last quarter of the sample for Category B and 189 days for Category A). Perhaps this is related to World Bank bureaucratic reforms initiated after 2016.

It may be that category A and B projects look different for more reasons than environmental impact. Certainly, policy lending projects are meant to look different.  Nonetheless, the results are at least suggestive. And more evidence for that comes from looking at the share of projects rated each category over time, presented in the figure below. It is very clear that Bank staff and borrowers are not keen to put forward projects that might be rated Category A: projects that involve considerable construction, for example. That’s going to be an issue if the World Bank wants to play a bigger role in creating low-carbon economies in middle income countries.

And the dramatic drop in category B projects in 2020-21 is also revealing: it reflects the World Bank’s efforts to get financing to countries suffering from the global pandemic.  The Bank reports that the average gap between completion of a project concept document and Board approval dropped from 10.6 months in 2019 to 7.8 months in 2021 (at a period during which annual IBRD and IDA commitments climbed from $45 billion to $67 billion). Clearly, World Bank staff and management appear to believe that if you want to get financing delivered fast, don’t trigger safeguards.

The results suggest the Bank should embrace large scale policy lending for climate. It is unlikely to trigger safeguards and is comparatively bureaucracy-free. That will make it more attractive to client countries, but also is likely to have a larger impact. Supporting policy change on issues such as carbon subsidies and pricing will have a larger effect on global emissions than financing the marginal infrastructure investment that may or may not have been low carbon without the World Bank’s financing.  

Still, when it comes to the bureaucracy of environmental review, it is time for a reassessment.  The World Bank Group should meet standards for environmental protection and social safeguards—it has financed some damaging projects with far too little in the way of remediation in the past. But the cost of the existing regime in terms of global climate and development outcomes may be too high. And that involves not just the review process but additional constraints including the ban on finance for nuclear power. If the Bank and other multilateral institutions are to play a greater role in climate finance, reforming their overly-bureaucratic investment approach through greater policy lending and a reformed safeguards regime needs to be part of the package—and that needs shareholder support.

Rights & Permissions

You may use and disseminate CGD’s publications under these conditions.