Analytics Report: Russian Economy 2020
Victoria Zaitceva
2024-03-19
Introdiction
Datasets
General Trends in the Russian Economy in 2020
Consumer Activity Trends:
-There was a discernible surge in consumer activity during the latter half of the year, reaching its peak in September.
-Conversely, April recorded the nadir of consumer activity, marking the lowest point in the year.
Online Payment Dynamics:
-Throughout 2020, the share of online payments in trade turnover demonstrated remarkable stability, maintaining a consistent 50% proportion across all months.
-This steadiness suggests a persistent reliance on online transactions amidst varying economic conditions and consumer behaviors.
Impact of COVID-19 Restrictions on Tourism:
-Analysis of domestic tourism trends in comparison to the preceding year (2019) unveiled a stark contrast, particularly evident during the spring holiday period of April and May. Notably, there was a significant decline, with the number of domestic tourists decreasing to half of the previous year’s figures.
-The decline can be attributed to the stringent anti-COVID measures in place across most regions, hindering mass tourism activities during this period.
Top Regions
Correlation analyses
No Connection Between Domestic Tourism and Non-Cash Payments:
The correlation analysis indicates that there is no discernible connection between fluctuations in domestic tourism levels and the prevalence of non-cash payments. This finding suggests that changes in domestic tourism do not significantly impact the share of non-cash transactions.
Relationship Between Loan Activity and Mortgage Transactions:
In regions where residents frequently apply for loans, there exists a notable correlation with higher volumes of mortgage transactions. This relationship demonstrates an average strength, with a correlation coefficient of 0.733. Such findings underscore the interplay between consumer borrowing behaviors and real estate market dynamics.
##
## Spearman's rank correlation rho
##
## data: as.numeric(top_only$Всего.ипотечных.сделок) and as.numeric(top_only$Всего.одобренных.заявок)
## S = 4183.1, p-value = 1.216e-10
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
## rho
## 0.7729539Weak Positive Relationship Between Online Payments and Online Loan Applications:
The correlation analysis reveals a modest positive relationship between the share of non-cash payments and the number of online loan applications, with a correlation coefficient of 0.366. While the association is weak, it suggests a tendency for increased online loan activity in regions with higher utilisation of non-cash payment methods.
The largest number of mortgage transactions (in total) in the most economically active regions occurs in the autumn.
Insights from Correlation Analysis
The correlation analysis conducted reveals a significant association between the frequency of loan applications by residents and the volume of mortgage transactions in various regions, characterized by a moderate strength of connection, with a Spearman correlation coefficient of 0.733. This suggests that regions with higher loan application rates also experience elevated levels of mortgage activity.
Seasonal Trends in Mortgage Transactions and Loan Applications
Echoing the observation of increased mortgage activity during autumn, the analysis indicates a similar trend in online loan applications. The lending activity during autumn can be attributed to several factors.
Autumnal Surge in Lending Activity
The surge in lending during the fall may be attributed to the post-summer holiday period and the pre-New Year phase, which often serves as an opportune time for individuals to make substantial financial decisions. Additionally, in the context of 2020, the relaxation of most anti-COVID measures by autumn might have alleviated decision-making barriers for individuals, facilitating increased financial activity.
Spring Dynamics in Online Applications
Conversely, the considerable variation in online loan applications observed across regions during spring can be attributed to the repercussions of COVID-19 restrictions. The introduction of stringent measures in March prompted individuals to defer loan applications due to heightened uncertainty. Those who did proceed with loan applications tended to favor online channels, given the closure of many physical bank branches.
People’s inclination towards online banking does not depend on the time
of year.
##
## Spearman's rank correlation rho
##
## data: top_only$Доля.онлайн.заявок.... and as.numeric(top_only$Сезон)
## S = 14315, p-value = 0.1276
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
## rho
## 0.2230163Insights from Visual Analysis
Visual analysis of mortgage transactions indicates a prevalent trend towards acquiring secondary housing during the winter and spring months. This observation is substantiated by a weak negative correlation, with a Spearman coefficient of -0.315, suggesting an increased incidence of secondary housing transactions at the onset of the year.
##
## Spearman's rank correlation rho
##
## data: top_only$Доля.сделок..вторичка.... and as.numeric(top_only$Сезон)
## S = 24243, p-value = 0.02875
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
## rho
## -0.3158496Potential Explanation for Seasonal Trends
The observed surge in secondary mortgage transactions during the winter and spring months may be attributed to following factor:
the slower pace of new home construction during colder seasons could limit the availability of newly-built properties, consequently driving up demand for existing secondary housing options. This phenomenon reflects a scenario where reduced supply intersects with sustained demand, prompting individuals to turn to the secondary housing market for viable options.
Regression analysis
When assessing the dependence of consumer activity and the share of non-cash payments on the average salary in the region, the number of active subscribers of wireless terrestrial fixed access to the Internet and the unemployment rate of the population, the following regression equations were obtained, respectively:
Y(activity) = 67.9 + 0.0001X(internet) - 0.486X(unempl) - 8.291I(spring) + 5.146I(summer) + 5.083I(aut)
Y(onl.trans) = 53.8 + 0.0001X(wage) - 1.313Х(unempl) + 1.046I(summer) + 3.157I(aut)
For consumer economic activity:
- About 41% of the changes in consumer activity can be explained by these factors.
- More internet users slightly increase consumer activity (by 0.0001 units for each additional user).
- A 1% rise in unemployment decreases consumer activity by 0.486 units.
- Compared to winter, consumer activity drops significantly in spring (by 8.291 units) but increases in summer and autumn (by 5.146 and 5.083 units, respectively).
For online payments:
- These factors explain around 55% of the variations in non-cash payments.
- A 1 ruble increase in average salary slightly boosts the use of non-cash payments (by 0.0001%).
- A 1% higher unemployment rate reduces the use of non-cash payments by 1.313%.
- The use of non-cash payments is notably higher in summer and autumn (by 1.046 and 3.157 units, respectively) compared to winter, with no significant change from winter to spring.
##
## Linear Regression Results
## ===========================================================================
## Dependent variable:
## ------------------------------
## Активность Безналичные_платежи
## (1) (2)
## ---------------------------------------------------------------------------
## Число.активных.абонентов.доступа.к.интернету 0.0001**
## (0.00004)
##
## Средняя.заработная.плата 0.0001***
## (0.00001)
##
## Уровень.безработицы -0.486*** -1.313***
## (0.076) (0.050)
##
## СезонВесна -8.291*** 0.640
## (0.645) (0.552)
##
## СезонЛето 5.146*** 1.046*
## (0.649) (0.554)
##
## СезонОсень 5.083*** 3.157***
## (0.648) (0.553)
##
## Constant 67.967*** 53.801***
## (0.669) (0.728)
##
## ---------------------------------------------------------------------------
## Observations 888 915
## R2 0.412 0.552
## Adjusted R2 0.409 0.549
## ===========================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01
##
## Linear Regression Results
## =====
## 0.750
## -----People’s preference for online banking and taking out mortgages is slightly influenced by the region’s average salary, internet subscribers, and unemployment rate.
z(online) = -7.60 + 0.0001X(wage) + 0.00004Х(internet) + 0.193Х(unempl) + 1.532I(spring) + 0.998I(aut)
z(mortgage) = 1.423 - 0.00001X(wage) + 0.0004Х(internet) - 0.256Х(unempl) + 0.827I(summer) + 1.829I(aut)
- Both online banking and mortgages are a bit more common with higher salaries and more internet users, but the impact is small.
- Unemployment has a bigger effect: a 1% higher unemployment rate means people are about 1.21 times more likely to use online banking. For mortgages, a 1% increase in unemployment makes people 1.29 times less likely to get one.
- Season matters too: people use online banking more in spring and autumn, and are more likely to take out mortgages in summer and autumn, compared to winter 2020.
##
## Logistic Regression Results
## ==================================================================================
## Dependent variable:
## -------------------------------------
## online_propensity mortgage_propensity
## (1) (2)
## ----------------------------------------------------------------------------------
## Средняя.заработная.плата 0.0001*** -0.00001***
## (0.00001) (0.00000)
##
## Число.активных.абонентов.доступа.к.интернету 0.00004** 0.0004***
## (0.00002) (0.00005)
##
## Уровень.безработицы 0.193*** -0.256***
## (0.033) (0.035)
##
## СезонВесна 1.532*** -0.312
## (0.353) (0.221)
##
## СезонЛето 0.105 0.827***
## (0.391) (0.225)
##
## СезонОсень 0.998*** 1.829***
## (0.362) (0.251)
##
## Constant -7.600*** 1.268***
## (0.597) (0.365)
##
## ----------------------------------------------------------------------------------
## Observations 915 915
## Log Likelihood -257.565 -452.098
## Akaike Inf. Crit. 529.130 918.195
## ==================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01
##
## Logistic Regression Results
## =====
## 0.750
## -----