Notes on the NZ Material Hardship Measures

Since 2008 Statistics New Zealand has measured material hardship in households with seventeen questions in the Household Economic Survey. This note reports on my brief exploration.[1]

The Unit of Measurement

The hardship questions are asked of a household and responses are reported on that base. The data reported below is by individuals so a household of four appears four times in some of the data. The questions reflect the household as a whole rather than the individuals in it. For instance, children in households that are forgoing dental treatment may be receiving full treatment in the free dental service but their parents may report they are deferring their dental treatment.

More generally, the analysis should not be interpreted to imply that the hardship is shared equally in the household. This may not be true; for instance, parents – especially mothers – may make sacrifices to shield their children from some of the hardship rigours the household faces.

Children in this paper are under 18 years of age. It is possible that households distinguished by the different ages of their children may show different patterns.

The Household Economic Survey does not cover the 1.7 percent New Zealanders who do not live in households but in institutions such prisons or rest homes.

The data comes from the 2022/3 year. It is from a sample surveys and subject to sampling noise. The data from the elderly is less reliable than all people or children. The adult results reported below are inferred as residual after deducting the other categories. Adult parents assigned to households with children assume the same pattern for one- and two- adult households.

The Questions

Some remarks about the individual questions listed.

The questions are similar to those asked in overseas surveys. They have to be generally applicable to almost every household. Hence there are no questions on period poverty since it is not a potential issue in most households. Even so, some of the questions are not applicable to, say, a vegetarian household and are asked in a way which recognises this.

The Household Economic Survey collects information on income and expenditure and also asks a question on income adequacy.

‘I would like you to think about how well your (you and your partner’s combined) total income meets your everyday needs for such things as accommodation, food, clothing and other necessities. Would you say you have not enough money, only just enough money, enough money, or more than enough money ?’

There is also some data on housing quality (e.g., keeping warm, dampness and mould) and while there is no direct questions about overcrowding it is possible to make some assessment based on the Survey.

However, currently it has not been possible to link these responses to the hardship measures. It is regrettable that it has not been done. I do not blame the statisticians in the government service who are already working hard. The failure reflects the inadequacies of our social science research funding. I tried to get funding to do similar research forty years ago; funding accessibility in the poverty area has not got any better since.

One of the first things proper investigation is likely to do is develop better hardship measures. This paper makes a small (unfunded) contribution.

The survey has been mainly used for assessing deprivation among children. It does not reflect the pressures on the elderly; for instance it does not cover an elderly person who is on a waiting list for hip surgery because they cannot afford to go privately.

More generally, the questions being asked are almost entirely about day-to-day expenses (perhaps month-to-month given some of the payment cycles). They do not cover big occasional expenses – like hip surgery. An inability to meet them may generate as much daily deprivation as those covered by the survey. [2]

Ranking the Hardship Responses

The responses to the hardship questions are ranked here by the proportion of responses in the population which indicate hardship. (The order is hardly different if the household responses, rather than people responses, are used.)

The order is from most common to the least common. The ranking can be interpreted as follows:

The first thing a typical struggling household does is to give up routine dental care (22.7% of people live in households which are forgoing dental care because of cost pressures). Next, it exhausts all its savings so that it has no emergency reserve of even $500 (16.5% of people). At the bottom of the list of seventeen responses, the most desperate households go without meat and fish three or more days a week (1.6% of people).[3]

Table 1 shows the results for all those in households, for those in households with children, for those in households with at least one person over 65, and for other adults. [4]

I grouped the responses into three. The ‘hardship’ category is of the more common responses (1-6) of ‘moderate hardship’. There is a noticeable step down to ‘severe hardship’ (7-14). The bottom three responses (15-17) are a ‘very severe hardship’ category, although there is not same drop between 14 and 15 as there is between 6 and 7. This is to emphasise the five-fold difference between responses to question 7 and question 17.

Table 1. Ranking by Hardship Responses

<> Rank   Household type People Children & Parents Elderly Adults only     % % % %   Hardship         1 Put off visits to dentists 22.7 28.0 12.0 21.2 2 Cannot afford unexpected $500 expense 16.5 22.1 9.0 13.3 3 Buy cheaper or less meat 15.8 20.8 9.0 13.0 4 Don’t have home contents insurance 13.9 18.0 7.0 12.2 5 Limited buying clothes/shoes 12.9 18.8 6.0 13.3 6 Cut back on local trips 12.2 16.0 6.6 10.2   Severe Hardship         7 Delay replacing/repairing appliances 8.2 11.5 3.4 6.3 8 Can’t borrow to meet costs 7.9 11.6 1.9 6.3 9 Put off doctor’s visits 7.4 9.8 2.1 6.8 10 Cannot pay utilities 6.0 10.3 1.9 2.7 11 Cannot buy fresh fruit or vegetables 5.9 7.6 2.5 5.2 12 Put up with feeling cold 5.8 6.7 3.1 5.9 13 Cannot pay for car 5.0 8.1 1.2 3.2 14 Unable to give gifts 4.7 5.7 3.3 4.1   Very Severe Hardship         15 Don’t have suitable clothes 3.0 4.2 1.1 2.4 16 Don’t have good shoes 2.0 2.9 0.4 1.6 17 Meals without meat, fish or chicken 1.6 1.7 0.6 1.7

Based on Responses to 2022/3 Household Economic Survey

The rankings for the sub-populations do not quite correspond to those for the total,. The differences may reflect sampling noise. Perhaps some can be explained by differences in living situations such as many of the elderly are unable to use or do not need to use a car.

There is a persistent pattern in which households with children show the highest responses to the deprivation questions and those with elderly the least.

Aggregating the Hardship Responses

Current practice (the Dep-17 score) is to add the number of hardship responses of each household, resulting in a categorical scale from 1 response to 12+ responses. We might assume that households make sacrifices in the same order. The generalisation seems to be roughly true in an aggregate tabulation, but I have yet to see a study which looks at individual households (that is, assesses how often there are exceptions to the rule). Hints in Table 4 below.

However, it is not obvious that each sacrifice is of equal weight. The relative weight could be assessed by a micro-analysis, although that would involve numerous heroic assumptions. Alternately, an aggregate approach would be to map responses against some external indicator. Possible indicators would be household income or, if they were recorded, the household’s or some independent observer’s assessment of the overall degree of hardship. (Mapping these hardship responses against income should result in much better household equivalence scales at the bottom of the income distribution instead of the current ones based on arbitrary judgements.)

Table 2. Scores of Hardship Responses

<>   Some Hardship   1 Put off visits to dentists 1.00 2 Cannot afford unexpected $500 expense 1.38 3 Cannot buy cheaper or less meat 1.44 4 Limited buying clothes/shoes 1.77 5 Don’t have home contents insurance 1.63 6 Cut back on local trips 1.86   Severe Hardship   7 Delay replacing/repairing appliances 2.79 8 Cannot borrow to meet costs 2.87 9 Put off doctor’s visits 3.08 10 Cannot pay utilities 3.81 11 Cannot buy fresh fruit or vegetables 3.88 12 Put up feeling cold 3.93 13 Could not pay for car 4.51 14 Unable to give gifts 4.88   Very Severe Hardship   15 Don’t have suitable clothes 7.58 16 Don’t have good shoes 11.44 17 Meals without meat, fish or chicken 14.63   MEDIAN 3.08

Calculated from Table 1 as explained in text.

Not having either the data base nor the means to do this, I used a more heuristic index to aggregate the data. It sets the score for the most common hardship (putting off visits to the dentist) at 1.0 (unity). All other reported hardships are scored by their infrequency relative to the most common hardship. Thus going without meat, fish or chicken for three meals or meals a week which occurs as .068 times as often as putting off the dentist is scored at 14.63 (1/.068 = 14.63). The larger the score the less frequently the hardship occurs (and, presumably, the more onerous it is). Table 2 (previous page) shows the ranking.

Applying these scores to the various categories we get the following average hardship scores among households in hardship. (The scores may appear low because they include households which report no deprivation responses.)

Table 3. Average Hardship Scores.

People                                       3.9

Households with Children                  5.3

Households with Elderly                     1.7     

Other Households                                3.1

Calculated from Table 1 as explained in text.

Not surprisingly, households with children are in greater hardship than households without children.

The average responses weighted by score were much the same for each type of household, so Table 4 shows only those for all people are shown here.

Table 4. Average Hardship Scores by Numbers of Household Hardships

<>  Responses People Min score  Ratios 1 1.8 1.0 1.78 2 4.1 2.4 1.72 3 6.7 3.8 1.77 4 9.5  5.4 1.74 5 12.7  7.2 1.76 6 15.9  9.1 1.75 7 18.8 11.9 1.58 8 21.9 14.7 1.49 9 28.2 17.8 1.58 10 30.3 21.6 1.40 11 34.9 25.5 1.37 12+ 46.9  29.4+  

Calculated from Table 1 as explained in text.

The result is not surprising. The people scores rise with increased number of hardship responses, rising at a near constant rate of 3.3 points between those households which report 1 hardship response and those which report 8 hardship responses. However the change from 8 responses to 9 responses exceeds 6 points which is a much bigger jump. The change is large enough to suggest that households with 9 or more hardship responses are struggling markedly harder than those below. Inspection shows that 9 is about the stage when households more frequently begin to cut back on eating meat, fish and chicken.

The conventional measure of household hardship is those households with 6 or more hardship responses. That is a judgement. It is not evident from this data that 6 is a more appropriate division than, say, 5 or 7. (Among those living in households with 6 or more hardship responses are about 145,000 children. A BHC50 income poverty line has about 150,000 children in households living below it. If there were an exact mapping we might equate the two as the same ‘poverty point’.)

The third column in Table 3 shows the lowest possible score for each category. For instance, the lowest possible score for households with only one hardship response would be 1.0. However only 31 percent of those households put off dental visits. Some have instead insufficient savings to meet an unexpected expense (14%), or have no home contents insurance (12%), or buy less or cheaper meat (12%) and so on. In the fourth column is the ratio of the reported scores with the lowest possible score. It comes to 1.8 giving some indication of the variation of choices.

(There is no judgement of poor housekeeping from this ratio. It is the result of household decisions reflecting their priorities and judgements. It is also possible that the household thought it easier to run down its savings and buy cheaper or less meat than give up a dentist, thereby ending up among the two-hardship-response households.)

For low hardship response households the ratio up to 6 hovers in the 1.7-1.8 range. For households in higher hardship, the ratio begins falling off, indicating that those households which are suffering greater hardship increasingly take the same measures.


There are no unexpected insights in this note. Its purpose was to consolidated my understandings of households which are economically struggling and quantified some of the magnitudes.

The tabulations confirm that on average households with children experience the most deprivation on these measures, although there are also other households which are struggling. Households with elderly in them appear to be struggling least, presumably as a consequences of the universal minimum income regime for the elderly.

However, the 17 questions do not cover some important dimensions of deprivation – especially housing. Nor do they deal with occasional but big expenditures without which there can be considerable hardship.

More generally, the paper shows that a scaling of responses rather than just counting of them provides some useful insights. The scale used here is not the best potentially available – it was that which could be derived from the available data. The construction of an even more rigorous scale and then linking it to household income would seem to be a priority for advancing the use of the data.


[1] I am grateful for help from Ryan Sutcliffe and Michelle Feyen of Statistics NZ who are not responsible for any opinions or errors.

[2] Because such expenses are infrequent the Household Survey is not really suitable to investigate them.

[3] That portion excludes those dieting or vegetarians and the like.

[4] The estimates for other households without children (i.e. adults only) are not direct but calculated by subtracting household with children and the elderly from the entire population.