Contents
In the mid-19th century, the German economist Ernst Engel formulated an important pattern of consumer behaviour: rising household incomes push down the share of food in household spending. As humans gradually cover their basic needs, they increasingly spend more on durable goods and services: ‘The poorer is a family, the greater is the proportion of the total outgo [family expenditures] which must be used for food. … The proportion of the outgo used for food, other things being equal is the best measure of the material standard of living of a population.’ Subsequently, household age and other characteristics were proposed as factors, but the model essentially remained the same.
Engel’s law at the macrolevel is evident in most countries, but not in Russia (link in Russian).
Between 2005 and 2023, Russia almost doubled its standard of living (also referred to as household consumption), measured as average real spending (adjusted for price changes). However, its growth was not even. At times, it was interrupted by crisis episodes, followed by recovery: the global financial crisis of 2008, the economic crisis of 2014–2016, the 2020 pandemic and the 2022 sanctions.
Yet, in defiance of Engel’s law, the share of household spending on food, including eating out, stagnated over the period: 30.4% in 2005 vs 29.7% in 2023 (up to 35% in the mid-2010s).
This phenomenon, referred to as the Russian consumption paradox, can be explained by several factors:
- Accelerating growth of food prices
- Increasing availability of ‘new’ goods and services
- Income and spending inequality
- Rising tax burden on households
Accelerating growth of food prices
The twofold increase in households’ welfare over almost two decades might have been expected to change consumer spending patterns. But that did not come to pass.
Behind the invariant percentage of spending on food is a combination of volume and price dynamics. In 2005–2023, the volume of average per capita food consumption grew 1.57 times compared to 1.99 times for the total value of household consumption. At the same time, average food prices grew 4.21 times, with prices for all goods and services growing only 3.41 times.
These volume and price movements led to the value of spending on food and all goods and services rising at nearly the same rate. As a result, the share of food in the structure of consumption changed little, if at all.
This is quite a credible explanation (link in Russian) as to why Engel’s law does not hold, because it assumes the stability of relative prices and consumer preferences, which is hardly a realistic assumption over a two-decade period.
Growing consumption of ‘new’ goods and services
A twofold increase in the volume of total per capita household consumption may be a statistical artefact: its calculation is methodologically correct, but there are questions about its economic interpretation.
According to Rosstat, between 2005 and 2023 the sharpest growth in consumption was seen in the Organization of Recreation, Sports and Cultural Events category – rising sevenfold.
This does not, however, suggest a sevenfold increase in theatre or museum visits, for example. Crucially, this is a very diverse group (link in Russian) of goods and services: in addition to visits to theatres, museums and stadiums, it includes, among others, purchases of computers and other electronic equipment, which have become much more affordable over the past 20 years. This is the component responsible for the sevenfold increase in the volume of consumption in this category.
Similar trends are found in the Information and Communication category, for which average per capita consumption rose 3.6 times. This growth was enabled by the expansion of internet and mobile services.
It was technological progress (1, 2 – links in Russian) that drove total consumption so much higher. Goods that were considered prohibitively expensive and niche in the mid-2000s are now viewed as part of a daily toolset. If the volume of consumption in these two categories had not changed, the overall growth of real consumption would have been 61% instead of the actual 99% (which is still considerable).
Consumption of food and non-alcoholic beverages was up only 1.54 times in 2005–2023. Some categories of consumption hardly grew at all. Alcohol and tobacco, for example, remained virtually flat over twenty years. They were up 1.5 times in the first half of the period under study, but virtually returned to initial levels thereafter.
All the other shifts in the cost structure proved rather limited. For example, share of spending on clothing and footwear went down from 7.6% to 5.4%; healthcare spending share increased from 2.9% to 7%. Other categories remained steady overall.
This means that the twofold increase in household consumption reflects not only higher incomes, but also expanding access to new types of goods and services.
Income and spending inequality
Despite some positive trends in the 2000s, these years were marked by extremely high income inequality. While the least wealthy 10% of households (first decile) spend more than half of their budget on food, the wealthiest 10% (tenth decile) spend about 20%. The difference is huge, but what is surprising is that food remains a large budget line even for wealthy households.
In the classic model, wealthy households would be expected to shift their consumption much more towards services, education, leisure, and investments in quality of life. In Russia, this shift is extremely slow.
In judging the inequality of incomes and spending among Russians, we must account for the limitations of official statistics. National accounts collect information only on households’ total spending, without breakdowns by income group. The main source of data on the structure of household expenditure is Rosstat’s Household Budget Survey (HBS). Due to its structure, it is inevitably characterised by both sampling error and systematic errors (1– link in Russian, 2). First, HBS is a poll that offers a limited dataset on households, and does not track changes over time; second, it rarely captures the wealthiest and poorest households.
Therefore, its results are sometimes drastically different from total household consumption statistics, which are based on direct measurements. The most striking example of this discrepancy is spending on alcoholic beverages and tobacco products. According to Rosstat’s Household Budget Survey, these goods accounted for only 2% of all consumer spending in 2023, compared with 7% according to household consumption statistics. Why is that? Some respondents are embarrassed to admit to buying alcohol and tobacco, and some cannot be included in the sample for objective reasons. As a result, total consumer spending in 2023 is estimated at ₽46 trillion in Rosstat’s household budget survey and at ₽73 trillion according to household consumption statistics.
To compare these two databases, we use an original algorithm based on the iterative proportional balancing (RAS) method. According to Rosstat’s Household Budget Survey, in 2023, the 10th decile group (the wealthiest) accounted for 28% of total consumer spending. However, the distribution of spending becomes much more uneven once the data are balanced. The share of the 10th decile group increases by 9pp to 37%, while the shares of the first to eighth decile groups decrease. After balancing, the ratio of spending of the tenth and first (least wealthy) decile groups increases from 9 times (according to Rosstat’s Household Budget Survey) to 20 times.
This adjustment of official statistics shows that actual spending inequality, and thus income inequality, is expectedly higher than indicated by Rosstat surveys. Overall, today’s economy increasingly faces the problem of incomplete data, especially when it comes to consumption by wealthy groups.
Rising tax burden on households
Recent years have seen a steady trend towards an increase in the tax burden (link in Russian) on households:
- The standard value added tax (VAT) rate was increased from 18% to 20% in 2019 and from 20% to 22% in 2026.
- The flat 13% personal income tax rate (link in Russian) after 2022 transformed into a more sophisticated differentiated scale (15% for ₽2.4–5 mln of annual income, 18% for ₽5–20 mln, 20% for ₽20–50 mln, and 22% for any amount above ₽50 mln).
- Personal income tax was put on bank deposits.
- Duties on excisable goods (alcohol, tobacco and fuel) were consistently increased.
- Customs duties for individuals were increased.
- Property tax base was increased.
- Recycling fees were consistently raised.
These fiscal innovations are affecting the current mechanism of resource redistribution between households, businesses and the state. They raised the total tax burden on households from 19.4% in 2016 to 23–24% in 2020–2024.
The changes in personal income tax and VAT are key determinants of the tax burden due to their magnitude. Changes in tax and VAT in 2025 and 2026, everything else being equal, lead to an increased share of low-income groups in the structure of aggregate demand.
In the case of personal income tax, this is because the increase only covered highly paid employees.
In the case of VAT, while its rise affects the whole population, the implications are stronger for wealthy groups, who spend most of their budgets on goods and services subject to a full tax rate. Low-income households have a higher share of spending on food and medicines, which are covered by preferential tax rates.
Our calculations show that the increase in the standard VAT rate in 2026 will raise the cost of goods and services consumed:
- by about 1.1% for the least wealthy groups;
- by about 1.8% for the wealthiest.
Also, the ‘everything else being equal’ assumption fails to account for potential adaptation of households to tax changes. For example, the wealthiest households, unlike low-income groups, may resort to their savings to sustain consumption.
The movements in relative spending of Russian households on food are inconsistent with trends across the globe. This is a marker of a persistent sense of financial ‘disorder’ felt by a significant part of the population, despite the almost twofold increase in the average level of wealth.
In assessing various socio-economic policies (levelling of incomes, poverty reduction, tax innovations, and food security) it is imperative to separately analyse the share of spending on food. This measure can sometimes be even more useful and informative than the average salary or average per capita income.
The full text of this research is published in journal Voprosy Ekonomiki, 2026, No. 5.
