Do sales events really raise consumption?
In their study of consumer response to the Great Singapore Sale, Agarwal, Koo and Qian (2022) show that substitution behaviours dampen overall consumption, and raise implications for the effectiveness of such mega-sales events.
24 March 2023
In a simple supply-demand model, when the price of a good is lowered, the quantity demanded increases. This premise underpins the logic behind sales events, where discounts are anticipated to attract elevated consumer spending.
In the real world, however, things are seldom so straightforward. In a complex marketplace filled with a diversity of substitute goods, consumers are continually assessing trade-offs between spending on a certain item but forgoing another, and at the same time also deciding whether to forsake inferior options for better alternatives when their purchasing power increases.
The Great Singapore Sale (GSS) is an annual, nationwide sales event launched in 1994 by the Singapore Tourism Board in alliance with the Singapore Retailers Association. While it was initially conceptualised as a means to promote Singapore as a shopper’s paradise and destination of choice for tourists in the region, the event was subsequently extended to two months and co-opted a broader range of merchants so as to engage domestic consumers as well. Discounts given during a GSS could go up to 70%, while products offered ranged from discretionary items such as watches and jewellery to necessities like F&B and apparel.
Will the GSS achieve its intended outcome of raising aggregate consumption (and thereby stimulate the economy), or, given the complex nature of the marketplace, will consumption behaviours temporarily alter but with no real increase in overall spending?
In their study “Consumption Response to Temporary Price Shock: Evidence from Singapore’s Annual Sale Event”, Agarwal, Koo and Qian (2022) investigate consumer behaviour before, during and after the GSS of 2010 to assess its impact on overall consumption.
Inter-temporal and cross-categorical substitution
Will a transient sales event have any long-lasting impact on a consumer’s propensity to consume? Intuitively, so long as disposable income remains the same, elevated expenditure in one area will naturally entail cuts in other areas. Consumption is therefore conceived as a function of disposable income – and this premise appears to pan out in real life.
In their study, Agarwal et al. found that while consumers spent more during the GSS, they reduced expenditure both before and after the sales event, exhibiting what is known in economic literature as “inter-temporal substitution”.
Additionally, consumers exhibited “cross-categorial substitution” by spending more on “shopping” category items that enjoyed GSS discounts and less on “non-shopping” category items that were unaffected by the sales event. The former includes goods like apparel, craft and hobbies, electronics, food and beverage, while the latter subsumed professional services, insurance, association memberships and other services unlikely to be related to the GSS.
In everyday terms, such behaviours are akin to cutting back expenses to save up for a big-ticket item, or returning to thriftiness after a major expenditure. And while fabulous discounts are on-going, why not seize the opportunity to make a purchase before prices go back to normal?
At the end of the period under analysis, any elevated consumption in response to sale discounts might have been netted off by relative austerity elsewhere, and when averaged out over time, there could be no real impact on overall consumption.
Empirical design
The researchers conducted their study based on a sample comprising over 120K debit and credit card holders from one of Singapore’s largest bank. Given the number of study subjects, and the ubiquitous reach of the bank, the sample was representative of the population at large. Data obtained from this sample included comprehensive transactional and demographic data, which allowed for heterogeneity tests whereby the researchers could investigate consumption behaviours across different subgroups of subjects.
Agarwal et al. focused on the GSS of 2010, which lasted between 27 May 2010 and 25 July 2010. That GSS was selected for study because in that year, the Singapore government announced a “Growth Dividend” payable in cash to Singaporean adults in the year following. Additionally, the Government also topped up the medical pension balance. Such a wealth effect was anticipated to amplify consumption behaviours and make changes in behavioural patterns more pronounced. The 8½ weeks of the GSS constituted the test period.
The reference period covered 4 to 2 weeks out from the GSS, and again, 2 to 4 weeks following the sales event, collectively making a total of 6 weeks. These 6 weeks served as the benchmark against which spending during the GSS was compared.
The researchers also set aside a week immediately prior to and after the GSS to “capture the immediate inter-temporal substitution effect”, or, in other words, to see if expenditures changed immediately before and after the sales event. The entire study period therefore spanned a stretch of approximately 16 weeks, from 29 April 2010 to 22 August 2010.
Merchant category codes included in the bank dataset enabled the researchers to distinguish between expenditure on goods highly likely to be affected by the GSS (i.e. shopping category items) and those unlikely to be affected (non-shopping category items). Such a distinction facilitated tests for cross-categorical substitution.
Findings
On the overall, total daily spending increased by about 55 cents during the GSS, or 2.9% compared to the daily average during the reference period. In the week following the GSS, the daily average dropped by nearly a dollar, or 4.8% relative to the reference period. There was, however, no significant change in the week prior to the sales event.
Changes in behavioural patterns, however, became clearer once the researchers drilled down to shopping-category spending. Consumers showed a dip of 34.5 cents (or -3.0%) in the week prior to the GSS, increased their spending by 85.4 cents (7.4%) during the GSS, and cut their expenditure by 44.5 cents (-3.9%) in the week after. Under this category of expenditure, it was clearer that consumers were exhibiting inter-temporal substitution by timing their purchases to coincide with generous discounts, and thereafter cutting back after a shopping spree.
In the non-shopping category, consumers were found to have reduced their expenses by about 30.6 cents per day (4.08%) during the GSS. This behaviour supports the consumption as a function of disposable income position, as money appeared to have been diverted away from this area towards the shopping category, with disposable income serving as a constraint on overall spending.
The findings therefore confirm Agarwal et al.’s expectation that consumers would show both inter-temporal substitution, spending less before the GSS and more during the GSS, and cross-categorical substitution, spending more on items discounted during the GSS and less on those unaffected by the GSS.
In dollar terms, the researchers estimated that total spending increased by about S$73 million during the GSS. However, this gain was netted off by reductions – during pre- and post-event periods, and spending cuts on non-shopping items during the GSS – which collectively amounted to approximately S$60 million. The net effect of the GSS had therefore been substantially dampened by inter-temporal and cross-categorical substitution behaviours.
Policy implications
Heterogeneity tests conducted by Agarwal et al. found that consumers exhibited little substitution behaviour when they used credit cards or when they were liquidity constrained. In other words, sales events are most effective when targeted at these groups.
A large-scale GSS may therefore be too broad and blunt an instrument to achieve intended outcomes.
Authors:
Agarwal, Sumit is the Low Tuck Kwong Professor at the NUS Business School and professor in the departments of economics, finance and real estate at the National University of Singapore.
Koo, Kang Mo is an associate professor at the Department of Economics at Yonsei University, Mirae Campus.
Qian, Wenlan is the Ng Teng Fong Chair Professor in Real Estate and professor of finance and real estate at the NUS Business School.