Aid donors used to primarily fund projects, but then they realised that broader economic issues in recipient countries derailed project success. This, along with other factors, encouraged them to shift their focus to structural adjustment. After structural adjustment largely failed, donors realised that underlying political institutions ‘ruled', and so they moved to providing budget support to gain influence in recipient countries where reform seemed possible. After this approach faltered, donors again turned to micro-level interventions, this time with a renewed focus on producing measurable results.

The above potted history asserts that change in aid modalities is driven by donor learning. It is a common story, and in The Development Dance Haley Swedlund suggests that it is – at minimum – highly incomplete. Rather than explain aid fads as the result of an ongoing learning process, Swedlund argues that cycles of aid delivery mechanisms result from donors and recipients being forced to work together to deliver aid while also being unable to sanction each other for breaking deals. The main issue is that while the low-level actors on both sides of an aid bargain may want to keep their word, both sides are ultimately accountable to actors other than each other. For example, a donor organisation might promise to deliver a set level of funding to certain sectors in a recipient country. However, the presence of this deal with the recipient government will not prevent the donor country's President from responding to electoral pressure by cutting the aid budget or shifting aid to new sectors. Similarly, recipients may pledge certain reforms in order to receive aid, but if key constituencies in the recipient country are harmed by these reforms then their passage may be unlikely. Aid fads result from donors and recipients constantly searching for – and failing to find – ways to create durable aid bargains.

This argument is supported by qualitative case study research in Ghana, Rwanda, Tanzania and Uganda and a survey of over 100 high-level aid officials from 23 different agencies operating in 20 countries in Africa. The results support the theoretical argument, and highlight the value of qualitative interviews in a research area that often places an emphasis on statistical analysis. Through discussions with staff of donor organisations and recipient governments, it is made clear that the broader literature's focus on delegation as the main concern of aid delivery is misplaced. Swedlund puts this concern to the side and focuses instead on illustrating the challenges of sustaining aid bargains when neither side can credibly threaten to sanction the other. This focus on the sustainability of aid bargains is a contribution to the literature.

I have one main criticism of the book, though my criticism is a request for an extension rather than a critique of the book's content. The book tells a consistent story across the case studies and the survey of aid officials. Perhaps there really is one underlying force consistently shaping aid bargains, but I could not help but wonder if there was also important heterogeneity. The book traces the rise and fall of budget support, but budget support did not rise and fall evenly for all donors and all recipients. We may have learned more about the institutions of aid bargaining if this heterogeneity was examined more seriously.

I expect that this book will be useful for academics, students and practitioners. It provides a political explanation to a question that has been unduly ignored. In emphasising aid bargains and credible commitments, it helps us understand the forces shaping how aid is delivered. It also allows us to tell a theoretically driven story of the history of aid modalities, and in doing so it will hopefully reduce our dependence on the potted history presented in my introduction.