Purpose For a public servant (PS) to be innovative, he or she needs to gather and process enough vital information from budget setting processes. However, research addressing how budgetary participation (BP) can trigger innovative behaviour (IB) in PSs and eventually foster task performance (TP) is rare, which is why the authors conduct this research. The purpose of this study is to understand how BP shapes TP through the IB of PSs. Design/methodology/approach The authors develop and test a mediation model with 860 responses from public sector workers across 25 government agencies using the PLS-SEM technique of Smart PLS 4. Possible control factors were addressed for both the mediator and target-independent construct. In particular, the authors use sex, age and tenure as control factors for IB. Also, the authors use job satisfaction, job engagement and perceived fairness in the budgetary system as control factors for TP. Findings The authors find a favourable and significant relationship between BP and TP; BP and IB; and IB and TP. The authors also find that IB partially mediates the relationship between BP and TP, such that BP fosters TP through the innovativeness of PSs. This finding suggests that PSs who participate in budget preparation are able to innovate, which, in turn enable them to perform tasks effectively. Research limitations/implications The authors call on forthcoming researchers to test the mediation model in other public sector settings worldwide. They may also consider other variables that can possibly mediate the positive impacts of BP on TP. Practical implications Lessons are discussed for governments, human resources directors and managers, management accountants, budget officers, procurement officers and other public sector workers and consultants. Originality/value The authors show how BP fosters TP through the innovativeness of PSs, since there is much more to know in this regard. The authors also help to resolve the paradox of inconsistency in the BP–TP literature by using IB as a mediator.