In this work a simple model for biological CO₂-fixation by RuBisCO has been studied where the usual C₅-substrate has been replaced by the C₂-carbohydrate glycolaldehyde in its deprotonated form which is the simplest ion with the same functionality as the actual substrate. The model system was studied both experimentally using physical mass spectrometry and quantum chemical calculations. It was found that carboxylation happens without a barrier, but the reaction is not selective and forms both carboxylates and carbonates. However, when the model system is extended with a Mg²⁺-cation similar to how RuBisCO incorporates a divalent cation at the active site, calculations show that the reaction becomes selective to formation of the carboxylate. The reverse reaction was also studied, that is decarboxylation of tartronate semialdehyde and its isomers. Since tartronate semialdehyde is not commercial and attempts at synthesizing it were unsuccessful, are arguments against the existence of the compound in bulk presented. Its isomer hydroxypyruvate was studied instead and the experimental barriers of decarboxylation agree well with calculations.