Social Media Activities & Retail Rent: Do influencers change the rent price?
Physical retail stores are increasingly under pressure: since 2010, the discussion of the “retail apocalypse” with the fast growth of e-commerce and closing of numerous stores. The COVID pandemic has unexpectedly accelerated the retail apocalypse. Not only because stores were forced to shut down, lockdowns also pushed people (especially the younger generation) to start and get used to online lifestyles. Around the same time, social media platforms, such as Facebook, Twitter, and Instagram, had “influencers”, who were paid to promote their brands and products to their followers. Influencers are social media users (not necessarily humans) who are paid to post advertising content (even if it does not appear to be “advertising”) in order to persuade other users to buy things.
Nowadays, this channel has become one of the most efficient advertising methods available, with the potential to save some retailers from closure. But what effect do influencers really have on the existence or physical retail stores, and can they really mitigate the declining brick-and-mortar shops?
This research on the impact of influencers on commercial real estate, startedin late 2018, when influencer was still a new thing in the US, and Instagram still had a user-friendly API for me to get data. I assessed whether if influencers are truly effective (and people believe so), by proving a relationship between influencers geographical location and the rent price of retail property.
In order to assess the effect of influencers on the retail rent prices, influencers must first be identified. The magic of influencers is based on the network nature of social media. A celebrity’s tweet spreads throughout the network and gets to millions of potential buyers. Meanwhile, an Instagram post about a cool art gallery from one of your closest friends will only reach a few dozens of people. How can we quantify influencers? In order to do this, I applied network theory to the Instagram and Twitter data for a large sample. Each ‘influencer score’ is determined by three centrality factors shown in figure 1: how many followers do they have (out-degree), how important are their followers (e.g. how many followers do they have; eigen-centrality), and how many messages are relayed by them (e.g. how central is he/she in their network; betweenness). As a result, in New York city, we identified 3606 influencers, mostly clustered in Manhattan (see figure 2).
Figure 1 - Graphic Theory Analysis
Note. We use 3 types of centrality measures to evaluate the relative “importance” of an influencer
In order to predict the effect of these influencers on retail rent prices, we first must acknowledge the spatial “spillover effects”. The effect of an influencer is not limited to their own house, but also to their neighborhood, where they most likely reside and post from frequently. In order to correct for this spatial effect, we cap the range by applying a dynamic bandwidth to the effect of each influencer: the importance (e.g. level) of influencer determines the reach of their individual effect. We additionally correct the rent prices for the neighborhood spillover effects (the rent of your neighbor will influence your rent).
Figure 2 - Geograpahical location of Influencer in our sample
Note. We identified 3606 influencer and their located posts.
Figure 3 shows the distribution of the coefficient estimation of influencer effect on New York City (focusing on Manhattan). The value of the coefficient ranges from -0.23 (dark red) to 0.46 (dark blue). Our first results show unequivocally that our influencers influence the retail rent prices, when correcting for neighborhood. Overall, the relationship is positive: in areas where influencers post, retail prices are significantly higher compared to areas where these influencers do not post.
Figure 3 - Geographical Coefficients Plot
Note. The effect coefficient of influencers, estimated using geographic weighted regression. The value of the coefficient ranges from -0.23 (dark red) to 0.46 (dark blue).
Interestingly, some defined sub-clusters show an opposite relationship: for East Midtown, Lower Manhattan, and Wall Street the presence of influencers “decreases” the retail rent prices. These clusters are yet to be explained. One hypothesis is that influencers work better in areas where young consumers are going. Another possible explanation for this spatial distribution is that the influencer effect on consumers varies across different retail sectors. For example, it is a reasonable guess that influencers are more effective for fashion stores than luxuries or durable goods. If we use Streetview from Google Maps and search the “red” zones in the coefficient map, (for instance, north of SOHO district) we can find clusters of luxury stores. It could well be that the rent price for these stores are high regardless of influencers.
Figure 4 - Google Streetview Comparison
Note: Street-view comparison between areas where influencers have positive (left side) and negative (right side) correlations with retail rent price.
However, we must be cautious in interpreting causality of our data. Influencers might be attracted by (areas with) successful retail stores, as oppose from retail stores seeking out influencers. It could also be a synergic correlation: successful retail stores attract influencers as much as influencers motivate retailers to start up shop (driving up rent prices). Regardless, these results shows that alternative data can be used to investigate novel relationships in real estate. For further investigation, or maybe retail acquisition orientation, social media influencer activity can be a valuable source of information.
Jin, S. Venus, Aziz Muqaddam, and Ehri Ryu. "Instafamous and social media influencer marketing." Marketing Intelligence & Planning (2019).
Pei, Sen, et al. "Searching for superspreaders of information in real-world social media." Scientific reports 4.1 (2014): 1-12.