The 4.2σ discrepancy between the standard model prediction for the muon anomalous magnetic moment aμ and the experimental result is accompanied by other anomalies. A crucial input for the prediction is the hadronic vacuum polarization {a}_{mu}^{mathrm{HVP}} inferred from σhad = σ(e+e−→ hadrons) data. However, the two most accurate determinations of σhad from KLOE and BaBar disagree by almost 3 σ. Additionally, the combined data-driven result disagrees with the most precise lattice determination of {a}_{mu}^{mathrm{HVP}} by 2.1 σ. We show that all these discrepancies could be accounted for by a new boson produced resonantly around the KLOE centre of mass energy and decaying promptly yielding e+e− and μ+μ− pairs in the final states. This gives rise to three different effects: (i) the additional e+e− events will affect the KLOE luminosity determination based on measurements of the Bhabha cross section, and in turn the inferred value of σhad; (ii) the additional μ+μ− events will affect the determination of σhad via the (luminosity independent) measurement of the ratio of π+π−γ versus μ+μ−γ events; (iii) loops involving the new boson would contribute directly to the prediction for aμ. We discuss in detail this possibility, and we present a simple model that can reconcile the KLOE and BaBar results for σhad, the data-driven and the lattice determinations of {a}_{mu}^{mathrm{HVP}} , the predicted and measured values of aμ, while complying with all phenomenological constraints.