Ongoing results
Daniel Nettle
2023-01-12
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 23-Aug.
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=0.62. 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=240) |
22-Oct (N=235) |
22-Nov (N=227) |
22-Dec (N=221) |
23-Jan (N=223) |
23-Feb (N=223) |
23-Mar (N=214) |
23-Apr (N=209) |
23-May (N=217) |
23-Jun (N=207) |
23-Jul (N=212) |
23-Aug (N=216) |
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| N-c income, reported | 1430 (1150) | 1460 (1210) | 1570 (1650) | 1380 (1200) | 1410 (1240) | 1460 (1330) | 1490 (1420) | 1570 (1480) | 1580 (1500) | 1540 (1430) | 1610 (1710) | 1530 (1320) |
| N-c income, calculated | 2210 (1850) | 2260 (1900) | 2260 (2130) | 2120 (1540) | 2190 (2100) | 2140 (2000) | 2070 (1550) | 2170 (2030) | 2260 (2070) | 2200 (1580) | 2280 (1960) | 2220 (2030) |
| Rent/mortgage | 673 (536) | 658 (521) | 657 (551) | 667 (562) | 658 (519) | 675 (556) | 668 (544) | 692 (631) | 682 (598) | 675 (553) | 656 (535) | 676 (575) |
| Water | 38.8 (34.0) | 40.8 (36.8) | 38.3 (27.5) | 42.5 (41.9) | 37.6 (27.4) | 37.0 (27.4) | 43.9 (42.8) | 41.3 (34.0) | 41.6 (34.9) | 42.6 (33.0) | 47.5 (45.4) | 44.2 (37.0) |
| Energy | 149 (94.9) | 150 (95.3) | 156 (92.3) | 177 (107) | 180 (114) | 187 (129) | 175 (111) | 173 (94.0) | 166 (93.2) | 163 (107) | 161 (104) | 159 (91.2) |
| Managing financially | ||||||||||||
| Comfortable | 10 (4.2%) | 12 (5.1%) | 9 (4.0%) | 12 (5.4%) | 14 (6.3%) | 15 (6.7%) | 12 (5.6%) | 14 (6.7%) | 12 (5.5%) | 13 (6.3%) | 10 (4.7%) | 13 (6.0%) |
| Alright | 22 (9.2%) | 16 (6.8%) | 26 (11.5%) | 21 (9.5%) | 18 (8.1%) | 20 (9.0%) | 18 (8.4%) | 17 (8.1%) | 19 (8.8%) | 17 (8.2%) | 18 (8.5%) | 26 (12.0%) |
| Getting by | 51 (21.3%) | 49 (20.9%) | 45 (19.8%) | 50 (22.6%) | 44 (19.7%) | 40 (17.9%) | 47 (22.0%) | 42 (20.1%) | 44 (20.3%) | 44 (21.3%) | 46 (21.7%) | 41 (19.0%) |
| Quite difficult | 112 (46.7%) | 119 (50.6%) | 105 (46.3%) | 104 (47.1%) | 108 (48.4%) | 109 (48.9%) | 97 (45.3%) | 97 (46.4%) | 98 (45.2%) | 92 (44.4%) | 102 (48.1%) | 95 (44.0%) |
| Very difficult | 40 (16.7%) | 39 (16.6%) | 42 (18.5%) | 34 (15.4%) | 38 (17.0%) | 38 (17.0%) | 40 (18.7%) | 39 (18.7%) | 44 (20.3%) | 41 (19.8%) | 36 (17.0%) | 41 (19.0%) |
And here is the same table for France:
| 22-Sep (N=232) |
22-Oct (N=208) |
22-Nov (N=195) |
22-Dec (N=190) |
23-Jan (N=194) |
23-Feb (N=196) |
23-Mar (N=184) |
23-Apr (N=191) |
23-May (N=175) |
23-Jun (N=168) |
23-Jul (N=150) |
23-Aug (N=157) |
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| N-c income, reported | 1070 (1010) | 1060 (935) | 967 (901) | 1140 (1070) | 1130 (997) | 1100 (954) | 1200 (1410) | 1080 (1010) | 1110 (1010) | 1120 (962) | 1100 (967) | 1070 (1120) |
| N-c income, calculated | 1840 (1970) | 1810 (1720) | 1620 (1360) | 1730 (1220) | 1770 (1420) | 1840 (1650) | 1750 (1380) | 1740 (1360) | 1730 (1160) | 1820 (1530) | 1810 (1200) | 1710 (1150) |
| Rent/mortgage | 915 (770) | 913 (753) | 880 (712) | 939 (830) | 916 (851) | 889 (747) | 927 (831) | 943 (808) | 918 (763) | 883 (755) | 874 (792) | 906 (774) |
| Water | 46.9 (51.1) | 48.9 (57.4) | 49.9 (56.8) | 58.6 (80.4) | 57.7 (109) | 56.1 (90.2) | 55.4 (66.4) | 64.7 (90.8) | 51.4 (50.3) | 53.9 (58.4) | 55.1 (76.1) | 57.3 (76.2) |
| Energy | 132 (115) | 134 (115) | 170 (475) | 151 (104) | 162 (137) | 179 (209) | 171 (135) | 163 (195) | 155 (127) | 144 (100) | 142 (121) | 141 (99.2) |
| Managing financially | ||||||||||||
| Comfortable | 12 (5.2%) | 12 (5.8%) | 13 (6.7%) | 12 (6.3%) | 9 (4.6%) | 10 (5.1%) | 8 (4.3%) | 12 (6.3%) | 6 (3.4%) | 6 (3.6%) | 3 (2.0%) | 5 (3.2%) |
| Alright | 26 (11.2%) | 15 (7.2%) | 11 (5.6%) | 18 (9.5%) | 21 (10.8%) | 16 (8.2%) | 18 (9.8%) | 16 (8.4%) | 18 (10.3%) | 15 (8.9%) | 13 (8.7%) | 19 (12.1%) |
| Getting by | 75 (32.3%) | 80 (38.5%) | 70 (35.9%) | 51 (26.8%) | 60 (30.9%) | 59 (30.1%) | 53 (28.8%) | 59 (30.9%) | 55 (31.4%) | 50 (29.8%) | 44 (29.3%) | 49 (31.2%) |
| Quite difficult | 97 (41.8%) | 83 (39.9%) | 79 (40.5%) | 83 (43.7%) | 78 (40.2%) | 86 (43.9%) | 81 (44.0%) | 77 (40.3%) | 73 (41.7%) | 70 (41.7%) | 67 (44.7%) | 65 (41.4%) |
| Very difficult | 22 (9.5%) | 17 (8.2%) | 18 (9.2%) | 23 (12.1%) | 24 (12.4%) | 23 (11.7%) | 24 (13.0%) | 26 (13.6%) | 22 (12.6%) | 27 (16.1%) | 23 (15.3%) | 18 (11.5%) |
We also define a measure of subjective economic
risk. This is the extent to which people feel themselves to be at
risk of economic crisis (destitution, losing their home, losing their
job). This measured (arbitrarily) on a scale of 0-300.
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=0.86.
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=235) |
22-Oct (N=235) |
22-Nov (N=227) |
22-Dec (N=221) |
23-Jan (N=222) |
23-Feb (N=222) |
23-Mar (N=214) |
23-Apr (N=209) |
23-May (N=217) |
23-Jun (N=206) |
23-Jul (N=211) |
23-Aug (N=216) |
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GAD anxiety | ||||||||||||
| None | 122 (51.9%) | 119 (50.6%) | 125 (55.1%) | 130 (58.8%) | 123 (55.4%) | 125 (56.3%) | 129 (60.3%) | 122 (58.4%) | 122 (56.2%) | 115 (55.8%) | 119 (56.4%) | 125 (57.9%) |
| Mild | 60 (25.5%) | 63 (26.8%) | 56 (24.7%) | 46 (20.8%) | 52 (23.4%) | 50 (22.5%) | 42 (19.6%) | 42 (20.1%) | 51 (23.5%) | 49 (23.8%) | 51 (24.2%) | 47 (21.8%) |
| Moderate-severe | 53 (22.6%) | 53 (22.6%) | 46 (20.3%) | 45 (20.4%) | 47 (21.2%) | 47 (21.2%) | 43 (20.1%) | 45 (21.5%) | 44 (20.3%) | 42 (20.4%) | 41 (19.4%) | 44 (20.4%) |
| PHQ depression | ||||||||||||
| None | 122 (51.9%) | 119 (50.6%) | 125 (55.1%) | 130 (58.8%) | 123 (55.4%) | 125 (56.3%) | 129 (60.3%) | 122 (58.4%) | 122 (56.2%) | 115 (55.8%) | 119 (56.4%) | 125 (57.9%) |
| Mild | 60 (25.5%) | 63 (26.8%) | 56 (24.7%) | 46 (20.8%) | 52 (23.4%) | 50 (22.5%) | 42 (19.6%) | 42 (20.1%) | 51 (23.5%) | 49 (23.8%) | 51 (24.2%) | 47 (21.8%) |
| Moderate-severe | 53 (22.6%) | 53 (22.6%) | 46 (20.3%) | 45 (20.4%) | 47 (21.2%) | 47 (21.2%) | 43 (20.1%) | 45 (21.5%) | 44 (20.3%) | 42 (20.4%) | 41 (19.4%) | 44 (20.4%) |
| Time discounting | 2.87 (2.26) | 2.97 (2.23) | 3.04 (2.21) | 3.12 (2.18) | 3.11 (2.20) | 3.06 (2.11) | 3.06 (2.19) | 3.07 (2.13) | 3.13 (2.26) | 3.13 (2.21) | 3.01 (2.21) | 3.09 (2.26) |
| Risk preference | 2.16 (1.38) | 2.16 (1.45) | 2.15 (1.37) | 2.11 (1.49) | 2.13 (1.36) | 2.14 (1.52) | 2.11 (1.44) | 2.14 (1.49) | 2.15 (1.46) | 2.23 (1.55) | 2.11 (1.43) | 2.00 (1.32) |
| 22-Sep (N=227) |
22-Oct (N=207) |
22-Nov (N=191) |
22-Dec (N=186) |
23-Jan (N=191) |
23-Feb (N=193) |
23-Mar (N=184) |
23-Apr (N=186) |
23-May (N=175) |
23-Jun (N=168) |
23-Jul (N=150) |
23-Aug (N=154) |
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GAD anxiety | ||||||||||||
| None | 95 (41.9%) | 94 (45.4%) | 87 (45.5%) | 84 (45.2%) | 86 (45.0%) | 94 (48.7%) | 84 (45.7%) | 85 (45.7%) | 89 (50.9%) | 83 (49.4%) | 69 (46.0%) | 75 (48.7%) |
| Mild | 86 (37.9%) | 69 (33.3%) | 57 (29.8%) | 68 (36.6%) | 65 (34.0%) | 56 (29.0%) | 60 (32.6%) | 68 (36.6%) | 55 (31.4%) | 54 (32.1%) | 42 (28.0%) | 46 (29.9%) |
| Moderate-severe | 46 (20.3%) | 44 (21.3%) | 47 (24.6%) | 34 (18.3%) | 40 (20.9%) | 43 (22.3%) | 40 (21.7%) | 33 (17.7%) | 31 (17.7%) | 31 (18.5%) | 39 (26.0%) | 33 (21.4%) |
| PHQ depression | ||||||||||||
| None | 95 (41.9%) | 94 (45.4%) | 87 (45.5%) | 84 (45.2%) | 86 (45.0%) | 94 (48.7%) | 84 (45.7%) | 85 (45.7%) | 89 (50.9%) | 83 (49.4%) | 69 (46.0%) | 75 (48.7%) |
| Mild | 86 (37.9%) | 69 (33.3%) | 57 (29.8%) | 68 (36.6%) | 65 (34.0%) | 56 (29.0%) | 60 (32.6%) | 68 (36.6%) | 55 (31.4%) | 54 (32.1%) | 42 (28.0%) | 46 (29.9%) |
| Moderate-severe | 46 (20.3%) | 44 (21.3%) | 47 (24.6%) | 34 (18.3%) | 40 (20.9%) | 43 (22.3%) | 40 (21.7%) | 33 (17.7%) | 31 (17.7%) | 31 (18.5%) | 39 (26.0%) | 33 (21.4%) |
| Time discounting | 2.85 (2.46) | 3.36 (2.35) | 3.57 (2.18) | 3.30 (2.20) | 3.23 (2.14) | 3.49 (2.31) | 3.34 (2.26) | 3.33 (2.25) | 3.29 (2.19) | 3.37 (2.21) | 3.25 (2.20) | 3.25 (2.26) |
| Risk preference | 2.19 (1.55) | 2.24 (1.58) | 2.31 (1.59) | 2.33 (1.63) | 2.39 (1.55) | 2.35 (1.55) | 2.42 (1.58) | 2.40 (1.63) | 2.30 (1.57) | 2.47 (1.52) | 2.39 (1.52) | 2.36 (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 up as income goes down? In a longitudinal study such as this one, you can ask this question in two ways: first, in the sample overall, are lower incomes generally associated with higher depression scores; and, within individuals, does their personal depression score increase as their personal income goes down. 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.
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 when income is lower, the psychological variable
(e.g. depression) goes up, 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 less income were months where they felt
more anxious. We are interested in whether most of the grey go in the
same direction as the big blue line.
Here is the graph 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 that come sooner.
We can also examine how the self-reported economic risk variable
relates to mental health. For example, here is how economic risk relates
to the anxiety GAD score in each month and country. As you can see, as
people’s perception of being at risk economically increases, so does
their anxiety and depression.
And here is the same information for PHQ depression score
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.