A large and influential literature aims to measure consumers’ spending, saving and debt responses to income changes using questions of this kind. These studies aim to estimate how much consumers would spend out of each £1 change in income. Average estimates of this parameter vary hugely across studies in ways that are unlikely to be due to differences in individuals’ economic environments. They may instead reflect differences in the scenarios put to consumers, or in the way questions are worded.
This experiment was designed to explain the large differences in findings across studies through varying methods of eliciting individuals’ spending responses to hypothetical income windfalls. Innovation Panel respondents were assigned to different conditions asking about how they would respond if they were to receive a one-time payment of either £500 or £2500. The condition of amount was crossed with a second condition, varying the duration that the windfall payment would be spent over, either 3 or 12 months. For both conditions, respondents were equally allocated to conditions. Respondents were asked if they would spend more, less or the same based on the amount and duration, and if more or less, the amount more/less.
The controlling variables, which were crossed and equally allocated at the household-level, are on record o_hhsamp_ip.
ff_mpctreatment(1/2 each, allocation stratified by ff_mpcduration ff_mpcamount sampleorig ff_gridmodew15
1 = Two-part question on marginal propensity to consume
2 = Direct question on marginal propensity to consume
ff_mpcduration (1/2 each, allocation stratified by ff_incentw15 sampleorig ff_gridmodew15)
1 = 3 months
2 = 12 months
ff_mpcamount (1/2 each, allocation stratified by ff_mpcduration ff_mpcamount sampleorig ff_gridmodew15)
1 = £500
2 = £2500
The variables used for this experiment are in the file o_indresp_ip:
t1mpc1, t1mpc2inc, t1mpc2red, t1mpc3same, t2mpc1500, t2mpc12500, mpcsustain, save, saved, t2mpc21, t2mpc22, t2mpc23, t2mpc24
