Contents
Over recent years, online shopping has changed from a specific digital practice into a routine part of everyday consumption. For many, ordering clothes, household appliances, books, cosmetics or homewares through marketplaces is as ordinary as shopping at a supermarket or a mall.
This has entrenched an idea that is of clear concern for traditional retail: digital commerce is steadily crowding out physical retail as high-street stores lose out to digital platforms in their offers, the speed of search and the ease of purchase. The higher the share of online sales, the less economic room, the argument goes, should remain for brick-and-mortar retail.
This finding is indeed supported by aggregate data. According to Rosstat ( .xlsx) (link in Russian), internet sales accounted for 11.3% of total sales in 2023, 15.2% in 2024 and 17.1% in 2025. For comparison, the share of internet sales was ( .xlsx) (link in Russian) a mere 0.7% in 2014. Admittedly, the methodology for calculating this indicator has been reworked over time, so the extreme values should be compared with caution. However, the scale of the structural shift is evident: it took a decade for online to grow from a peripheral segment into a sizeable channel of Russian retail.
Having said this, the data on shoppers’ behaviour fail to support the simple view that every additional ruble of online sales is supposedly one ruble lost to physical stores. The third 2026 edition of the Household Economic Behaviour Barometer (link in Russian), an HSE University project, shows that in spring of 2026, 88.4% of respondents went to chain stores to buy non-food goods, and 44.2% turned to marketplaces. Only 29.4% of shoppers limited themselves to just one retail format; 37.5% used two, and a further 33.1% used three or more, such as a chain store and a marketplace or a chain store, a non-chain store and a marketplace. In other words, the spread of marketplaces does not so much entail the mass rejection of brick-and-mortar retail but an increasingly complicated consumer journey as online complements current shopping channels.
This finding is crucial, since it changes the economic statement of the problem. If marketplaces simply replace regular shops, we are looking at the technological redistribution of real demand: a consumer who wants to buy a pair of shoes now opens an app instead of going to a mall. This scenario may entail no significant change in total consumption. However, if the digital channel is used alongside brick-and-mortar retail, and particularly intensively by households that are big spenders, then this ‘platformisation’ potentially affects not only the choice of point of sale but of the spending mechanism itself. The substantive discussion is then not about the death of brick-and-mortar retail but whether the new institutional environment is expanding the volume of consumption that households consider normal and affordable.
The classics on income and consumption
The allocation of income between consumption and saving was the initial focus of a large part of classical consumer theory. John Maynard Keynes, who laid the foundations of modern macroeconomics, explains that, in theory, a rise in disposable income increases consumption, but not by the full amount of the rise: part of the additional income goes into savings. The increased consumption to increased income ratio is described by the marginal propensity to consume, i.e. the average willingness to buy something. Keynes demonstrates the significance of this indicator not only for individual households but for the broader economy: the greater the share of additional income that is returned to consumption, the more intensely an initial rise in spending spreads throughout the economy.
This can be described in simple terms as follows: if a family’s additional income is ₽10,000 and they spend ₽8,000 and save ₽2,000, the marginal propensity to consume is 0.8 – the share of the additional income that becomes additional demand. In The General Theory of Employment, Interest and Money, Keynes lists reasons why people refrain from spending, such as the need to save for unforeseen circumstances, to prepare for a future decline in income, to pay for children’s education or to provide for old age.
Subsequent research has challenged this dependence of current consumption on current income as too direct. In his permanent income hypothesis, Nobel laureate Milton Friedman identified permanent and a transitory components of income and showed that households respond to them differently. An individual receiving a one-off bonus in December, for instance, may save a significant part of it. If the same individual receives a permanent pay rise that leads them to expect an income that is higher still in one year, their current spending may rise much more. This is why Friedman linked consumption primarily to anticipated long-term income, rather than to than current income.
Nobel laureate Franco Modigliani and his colleague Richard Brumberg proposed a different – but still implicitly close – argument. In the life-cycle hypothesis, individuals allocate resources across the periods of their lives: working age individuals build up savings to maintain an acceptable level of consumption after retirement. This means that identical current incomes may come with different behaviour depending on age (whether an individual is 25 or 60 years old), differences in accumulated wealth, and expected future earnings or planning horizons. This logic has seen many instances of empirical application, including in the analysis of pension systems and the link between demographic structures and saving.
Russian background: high rates
In many approaches within classical consumer theory, consumption is linked to the choice between using resources today or in the future. This is why what matters in the Russian context is not only the spread of marketplaces but the unusually high cost of money during the recent expansion of digital platforms. In early 2024, the Bank of Russia’s key rate was 16%; in October 2024, it was raised to 21% and remained at that level through June 2025. Only then did a gradual easing cycle begin, taking it to 14% in September 2026. The past three years have seen not only high rates but unprecedentedly tight monetary conditions, which enable high returns on savings and constrain consumption on credit.
The working of this mechanism is fairly simple for ordinary consumers. If a deposit pays a high nominal return, consumers who choose not to buy anything receive a clear monetary benefit: ₽100,000, for example, will earn no interest if it is spent right away. At the same time, the high lending rates mean that buying on account of future income is less attractive. All else being equal, this environment should support saving and hold back discretionary spending.
It is therefore particularly striking that the rapid rise of digital commerce and multichannel consumer behaviour occurred precisely in a period when the traditional interest-rate mechanism was working in the opposite direction.
Digital platforms simplify the purchase process
While they do not remove budget constraints or create income out of nothing, digital platforms change the cost of the consumption process. As British economist and Nobel Prize winner Ronald Coase explains, market transactions are not free: the seller must be found, information obtained, the terms considered and the exchange arranged. American economist Oliver Williamson, also a Nobel laureate, develops the idea further, showing that both the structure of transactions and the ways they are organised have economic significance in their own right.
For consumers, marketplaces cut out a wide selection of these transaction costs: they no longer have to travel to several shops, check whether items are in stock, or compare prices ‘by-hand’. Buying a vacuum cleaner just 20 years ago required a trip to a shopping centre and visits to several shops. Today, dozens of models fit on a single screen for comparison. The reduction in the full cost of the purchase is real even if the price of the appliance itself is the same.
However, transaction costs provide only part of the explanation for the change in consumer behaviour. In the example above, consumers deciding beforehand to buy a vacuum cleaner can do so rapidly and essentially cheaply by turning to digital platforms. This does not, however, necessarily lead consumers to buy more goods overall. To explain this shift, it is necessary to explain the reasons for the changes not only in the methods of meeting existing needs but also in the set of needs that consumers consider worthy of spending on.
Digital platforms change what is seen as normal consumption
This is where the analysis of consumption comes closer to the
ideas of Thorstein Veblen, the founder of the institutional school in political economy, than to the fixed-preferences model. Veblen introduced the concept of ‘conspicuous consumption’ – the practice of buying expensive goods and services to display wealth and status.
His reasoning was broader than consumption merely for the sake of demonstration: as consumption standards take shape in a social environment, an established standard of living gradually becomes a norm to be maintained – and thus involves spending.
Veblen contended that consumption is linked to the comparison of oneself with others and with habits that take hold within the social environment. While a car of a certain class, regular restaurant meals or a certain type of clothing formerly served as visible markers of status, the digital environment today greatly expands both the space for comparison and the number of goods that make up the perceived standard of a normal life.
Digital platforms reinforce this mechanism not because every shopper necessarily seeks to display status to those around them but because they always present alternative standards of consumption. A consumer opens an app to buy just one specific item but finds dozens of varieties on display, as well as related goods, recommendations, ratings and personalised offers. A short while ago, food delivery was simply an added convenience, while today it may be seen as a standard way to free up time. A taxi ride instead of public transport is no longer seen as an exception. Next-day delivery gradually fosters the impression that a longer wait is a loss of convenience. The Veblenian mechanism here manifests itself not necessarily in ostentatious luxury but as a gradual change in the set of goods and services considered part of normal everyday life.
This idea has been further developed in later research on consumption in which one’s own income is no longer the sole reference-point. James Duesenberry, the creator of the relative income hypothesis, associates spending with an individual’s position relative to those around them and with the ‘demonstration effect’: exposure to a higher standard of consumption puts pressure on one’s own level of spending. Contemporary research also finds evidence that changes in the income of certain households can influence the behaviour of neighbours and social groups precisely through the comparison of standards of consumption (1, 2, 3).
In Russian economic sociology, there is a closely related strand of research represented by Vadim Radaev, head of the HSE Laboratory for Studies in Economic Sociology. Investigating non-standard consumption, he treats panic, impulsive, compulsive, status and ethical consumption not as rare deviations from ‘normal’ behaviour but as persistent practices that can be observed and measured based on mass data. Radaev finds that consumer decisions are often driven, in addition to the estimated economic benefit, by emotional, moral and status motives. This approach to analysing digital platforms matters precisely because the digital environment reduces the interval between the emergence of a motive and the ability to turn it into a transaction.
Russian data on multichannel consumption are well aligned with this logic, although they themselves do not confirm a causal effect from digital platforms. Fresh data from the HSE’s Household Economic Behaviour Barometer show that the complexity of consumer strategy is related to material means. As regards non-food products, the least well-off respondents are much more likely to limit themselves to a single shopping format, consumers in the intermediate group most often use two, while the better-off (those who can easily buy big-ticket household appliances, other than cars) tend to combine, more often than the others, three or four shopping formats. A similar relationship holds for food: the least well-off tend to rely on a single format, while the better-off use three.
The socio-demographic profile of multichannel shopping also matters. According to the Barometer, among buyers of non-food goods, younger respondents more often use three or more formats, but older groups tend to make do with just one. Higher education is also associated with a more complex combination of the channels. Surprisingly, though logically, the difference in the size of settlement is also relevant: when it comes to non-food shopping, small town residents tend to prefer multiple formats, while the residents of large cities more often confine themselves to one. This is fully consistent with the economic role of digital platforms as a means of widening the range available locally: in locations where physical retail already offers a wide range, the added value of one more channel may be lower than in a small town.
Active online shoppers spend more
Still more direct evidence comes from my own study of the relationship between household spending and online and offline purchases. It is based on representative microdata for the Russian population in 2025 and is to be published in Studies on Russian Economic Development (link in Russian) (2026, No. 6 November). In the working sample, 88.46% of respondents made online purchases at least once a month, and 92.94% shopped offline. The frequency with which consumers choose one channel or the other, however, differs markedly: 9.73% shop online daily, and 27.93% shop offline daily. This alone shows that the digital channel has yet to replace the traditional one, even given the very deep penetration of online shopping into everyday life.
The key finding, however, is not about the prevalence of the channels but their different correlations with spending. For respondents who shop online every day, median total monthly household spending is ₽70,000, while for those who do not shop online, it is ₽42,000. In the case of offline purchases, the values are ₽60,000 and ₽40,000. The difference in the mean spending between the extreme groups is even more telling: around ₽53,000 for online against roughly ₽30,000 for offline. The habitual level of spending thus confirms that the intensity of digital consumption is far more strongly differentiated by household spending level than is the intensity of traditional consumption.
At the same time, I find no signs of the simple substitution of one channel for the other. Activity in one channel is positively associated with the probability of intense shopping activity in the other, after controlling for sex, age, education, spending, size of settlement and other attributes. In other words, being a heavy online spender does not mean that one is an irregular offline shopper; on the contrary, intense activity often coexists across the two channels.
Further evidence of an expansion in the consumer space comes from the link between purchases of products and of other platform services. Of respondents who buy goods online every day, 85.2% use taxi hailing via the internet or an app, and 88.6% use food delivery services. The shares for those who make no online purchases are 31.5% and 9.6% respectively. This relationship is much weaker in the case of offline purchases. I read this as the emergence of a digital consumer ecosystem in which individual practices on digital platforms are mutually complementary.
These results do not make the case that a marketplace in itself propels households to spend more. A better-off and more active consumer may a priori use both shops and apps more often, so the direction of causality requires separate testing. Nevertheless, the combination of empirical results from flash data suggests that the inference that offline activity is being crowded out is inadequate. At the level of mass behaviour, what is observed is not an exit from one channel into another, but the emergence of new channels and services, with well-off groups tending to show the most complex consumer practices.
This is where I propose revisiting classical consumption theory from a somewhat different perspective. Keynes, Friedman and Modigliani made strong arguments about how current and expected income, savings, age and intertemporal choice determine the balance between spending and saving. While the digital platform economy does not undermine any of these mechanisms, it adds to them an institutional environment that both lowers the cost of making purchases and widens the options available to consumers. In the Veblenian tradition, there is another important consideration: the concept of the acceptable level of consumption does not necessarily remain constant, since it changes along with social norms and observed standards of living.
This may result in a mechanism that cannot be reduced to either lower prices or the emergence of new shops. The digital environment creates multiple ways to easily spend additional income. If a family has, say, ₽10,000 left after essential spending, the decision to save this much competes not only with planned big-ticket purchases but with a multitude of small and quickly accessible transactions: items from recommendations, extra deliveries, taxi rides, subscriptions, or other services whose full cost of acquisition is low thanks to the platform. This does not mean that every such opportunity will be taken, but the institutional ‘price of entry’ into additional consumption becomes lower.
It is particularly notable that this process has unfolded against a backdrop of tight monetary conditions. In 2024–2025, the high key rate raised the returns on savings, increasing the cost of consumption on credit, that is, it created an incentive not to spend part of current income. Concurrently, the digital infrastructure worked in the other direction, reducing the non-financial costs of purchases and widening the set of goods and services available. The resilience of online consumption and its complementarity to offline observed in such an environment do not prove that the platform effect is stronger than the interest-rate channel, but they do give grounds to include the effects of digital platforms in the analysis of consumer behaviour as a separate institutional factor.
Platformisation as driver of expansion of consumption
Therefore, the main macroeconomic conclusion should for now be a testable hypothesis rather than an established fact. The data available provide enough evidence that the changes in the Russian market are driven not only by the redistribution of sales between physical and digital sellers. Digital platforms reduce the search and transaction costs, widen the offer, make a number of services part of the everyday standard, and intensively embed themselves in the consumption of higher-spending households. The totality of these mechanisms may raise the chances of additional income going to consumption rather than to savings.
If this effect proves persistent and significant at the aggregate level, the rise of digital platforms may bring about a higher marginal propensity to consume.
Current surveys of the digital platform economy suggest a cautiously formulated hypothesis: today, digital platforms affect both the distribution of demand that is already present and also the probability that it is met through the personalised ranking, better matching of supply and demand, lower search costs and simplified customer journey of digital platforms. The findings of my article likewise allow only for this mechanism to be identified and for empirical facts consistent with it to be shown to support the conclusions of the aforementioned review. The next step will require much macroeconomic testing: panel data on income, spending and digital platform use, attention to interest-rate conditions, and strategies for identification that isolate the influence of the digital environment from the inherent differences between households.
From this perspective, the debate over whether online will crowd out offline shopping appears misplaced and unnecessary. Russian data rather point to the emergence of a market in which the digital and traditional channels coexist and the expansion of digital platforms widens the number of ways to complete transactions and spend money and raises the propensity to consume, even in an environment that strongly favours saving.
The most significant economic consequence of this transformation may be, rather than the end of brick-and-mortar retail, a change in the consumption function: given the same additional income and the same desire to save, the institutional environment offers ever simpler ways of immediately turning resources into purchases. This is precisely why the hypothesis of a rise in the marginal propensity to consume under the influence of digital platforms deserves separate macroeconomic testing, especially in an economy marked by the rapid digitalisation of retail and the exceptionally favourable saving conditions over the past two years.