The Changing Cost of Living study is an ongoing panel study in the UK and France, dealing with how people’s mental health changes as the cost of living puts a squeeze on their available income. The cross-sectional association between financial difficulties and mental health (people with less available money have poorer mental health) has been demonstrated many times. It’s much rarer to examine how people’s mental health changes in real time as their financial circumstances change. We started in September 2022 with a panel of 250 people in the UK and 250 in France. We are checking in with our panel members every month for the coming months.

On this webpage, we are producing interim summary stats as the study goes on. With time we may be able to see trends and patterns in the data. This page will auto-update. At the moment, it is using data collected up until 22-Nov.

Finances

Our main financial measure is non-committed income, which is income left over in the household after rent or mortgage, council tax (UK), water and energy bills are paid. We have two measures of this, one based on the direct report by the participant, and the other calculated by us by subtracting the various kinds of costs each month from the income received. The reported measure and the calculated measure are correlated at r=0.75. All income measures are equivalised (adjusted for household size) and converted to euros at purchasing power parity exchange rates (so that the same number means ability to buy the same things). We also ask participants how they are managing financially each month.

The table below shows how the various financial measures are distributed by month in the UK.

22-Sep
(N=245)
22-Oct
(N=235)
22-Nov
(N=227)
N-c income, reported 1440 (1150) 1470 (1210) 1460 (1190)
N-c income, calculated 2040 (1250) 2050 (1290) 2010 (1320)
Rent/mortgage 682 (543) 658 (521) 657 (551)
Water 38.8 (33.9) 40.8 (36.8) 38.3 (27.5)
Energy 149 (94.9) 150 (95.3) 156 (92.3)
Managing financially
Comfortable 10 (4.1%) 12 (5.1%) 9 (4.0%)
Alright 22 (9.0%) 16 (6.8%) 26 (11.5%)
Getting by 52 (21.2%) 49 (20.9%) 45 (19.8%)
Quite difficult 113 (46.1%) 119 (50.6%) 105 (46.3%)
Very difficult 40 (16.3%) 39 (16.6%) 42 (18.5%)


And here is the same table for France:

22-Sep
(N=246)
22-Oct
(N=209)
22-Nov
(N=189)
N-c income, reported 1080 (980) 1070 (1010) 979 (903)
N-c income, calculated 1560 (1010) 1620 (1200) 1440 (1350)
Rent/mortgage 908 (757) 919 (756) 910 (802)
Water 48.0 (55.1) 48.9 (57.3) 50.4 (57.6)
Energy 132 (113) 135 (115) 171 (478)
Managing financially
Comfortable 14 (5.7%) 12 (5.7%) 13 (6.9%)
Alright 30 (12.2%) 15 (7.2%) 11 (5.8%)
Getting by 77 (31.3%) 80 (38.3%) 69 (36.5%)
Quite difficult 102 (41.5%) 83 (39.7%) 78 (41.3%)
Very difficult 23 (9.3%) 18 (8.6%) 18 (9.5%)



Mental health measures, time and risk preference.

We use two main measures of mental health, an anxiety score (which is called the GAD score), and a depression score (which is called the PHQ score). These are scales developed to detect clinical levels of difficulty. For each respondent, they give both a continuous number, and a classification into no difficulties, mild difficulties, and moderate-severe difficulties. Anxiety and depression tend to go together: the correlation between the two scores is r=0.84.
We also have a measure of time preference (do what extent do you prefer a smaller, sooner reward over a larger, later one?) and risk preference (to what extent do you prefer a smaller sure payoff over a larger unsure one?). We don’t call these mental health measures since there is nothing necessarily unhealthy about preferring a smaller, sooner reward: it could be perfectly rational under some circumstances, and the same goes for risk preference.
The next tables show the distribution of anxiety, depression, time preference, and risk preference, by month, for the UK and for France respectively.

22-Sep
(N=237)
22-Oct
(N=235)
22-Nov
(N=227)
GAD anxiety
None 124 (52.3%) 119 (50.6%) 125 (55.1%)
Mild 60 (25.3%) 63 (26.8%) 56 (24.7%)
Moderate-severe 53 (22.4%) 53 (22.6%) 46 (20.3%)
PHQ depression
None 124 (52.3%) 119 (50.6%) 125 (55.1%)
Mild 60 (25.3%) 63 (26.8%) 56 (24.7%)
Moderate-severe 53 (22.4%) 53 (22.6%) 46 (20.3%)
Time discounting 2.88 (2.25) 2.97 (2.23) 3.04 (2.21)
Risk preference 2.16 (1.38) 2.16 (1.45) 2.15 (1.37)


22-Sep
(N=241)
22-Oct
(N=208)
22-Nov
(N=189)
GAD anxiety
None 98 (40.7%) 94 (45.2%) 85 (45.0%)
Mild 90 (37.3%) 69 (33.2%) 56 (29.6%)
Moderate-severe 53 (22.0%) 45 (21.6%) 48 (25.4%)
PHQ depression
None 98 (40.7%) 94 (45.2%) 85 (45.0%)
Mild 90 (37.3%) 69 (33.2%) 56 (29.6%)
Moderate-severe 53 (22.0%) 45 (21.6%) 48 (25.4%)
Time discounting 2.93 (2.48) 3.36 (2.34) 3.59 (2.20)
Risk preference 2.19 (1.51) 2.23 (1.58) 2.33 (1.60)


Relationships between income and psychological measures

The key question for the study is how psychological measures vary with available income. For example, do depression and anxiety go down as income goes up? In a longitudinal study such as this one, you can ask this question in two ways: first, in the sample overall, are higher incomes generally associated with lower depression scores; and, within individuals, does their personal depression score decline as their personal income increases. These two results are not identical. For example, in the UK, it is the case that higher incomes are associated with taller heights in the population overall, but it is not the case that when an adult’s personal income increases, they get taller.

We will be continuously studying this question as the study goes on. The graphs here show the relationships in the data so far - we expect these to change as we get more data from each individual. In the graphs, the blue line shows the relationship between non-committed income and the psychological variable in the population overall. If the blue line slopes down from left to right, that means that as income goes up, the psychological variable goes down, in the population as a whole. The little grey lines show the relationship for each individual in the study (remember, for each person, we measure their non-committed income repeatedly over time, and also their psychological variables repeatedly over time). So, again, a grey line that sloped down from left to right would mean for that one person, months with more income were months where they felt less anxious. We are interested in whether most of the grey lines will go in the same direction as the big blue line. Of course, they will not follow perfectly: people become more or less anxious for all kinds of reasons, not all of which have anything to do with money. But it will be interesting if generally speaking, more of the grey lines end up sloping one way than the other.The precision of the little grey lines will improve as we get more datapoints from each person.

Here is the graph so far for anxiety (a higher score means more anxiety).

And now for depression (a higher score means more depression):

And finally here is the graph for time discounting. Here a higher score means a stronger preference for payoffs than come sooner.

About the Changing Cost of Living Study

The Changing Cost of Living Study is funded by the University of York Cost of Living Research Group; the ActEarly Collaboration (UK Prevention Research Partnership), and the Agence Nationale de la Recherche (ANR-21-CE28-0009). The researchers are from the Ecole Normale Superieure, Northumbria University, and the University of York.