View Notebook

Time Series Analysis

Project Overview

This project presents a comprehensive analysis of a hypothetical brand's sales performance operating in the FMCG in relation to its marketing strategies and the corresponding competitive landscape.

Purpose of the Analysis:

The primary objective of this analysis is to evaluate the effectiveness of the brand's marketing initiatives and their interplay with competitor strategies. By scrutinizing sales, advertising, and pricing data, this report aims to empower the brand's management with actionable insights for data-driven decision-making. The analysis can be broken into:

  1. Analysis of the brand's marketing effectiveness

  2. Analysis of the brand's marketing interaction with competitors

  3. Analysis of the relative impact of the brand's own and competitors' marketing action on sales.

Outline

Part 1: Import dataset and necessary libraries

Part 2: Data preprocessing - data wrangling & cleaning

Part 3: Exploratory data analysis (EDA)

  • Descriptive Analysis

  • Temporal Causality Analysis

  • Stationarity Analysis

  • Dynamic System Analysis

  • Variance Decomposition Analysis

Part 4: Conclusion: manegerial insights & recommendations

Data Descriptions

The dataset contains 207 weekly observations of the brand's performance metrics including log-transformed sales, advertising expenditure, and pricing data, and competitors' actions including competitors' price and advertising.

  1. Week - Identifier for the week of observation; 1-207 (linear trend)

  2. LnSales - Log-transformed volume sales of the brand (expressed in the relevant unit)

  3. LnAdvertising - Log-transformed advertising spending by the brand (in dollars)

  4. LnPrice - Log-transformed unit price of the brand (expressed in dollars per unit)

  5. LnTotalCompAdvertising - Log-transformed total advertising spending by competitors of the brand (expressed in dollars)

  6. LnAvgCompPrice - Log-transformed average unit price of competitors of the brand (expressed in dollars per unit)

  7. Qrtr 1 - Indicator variable = 1 for the first quarter, 0 otherwise

  8. Qrtr 2 - Indicator variable = 1 for the second quarter, 0 otherwise

  9. Qrtr 3 - Indicator variable = 1 for the third quarter, 0 otherwise

  10. Qrtr 4 - Indicator variable = 1 for the fourth quarter, 0 otherwise

Part 1: Import Dataset & Libraries

Part 2: Data Preprocessing - Data Wrangling & Cleaning

Pre-processing involved verifying data integrity, handling anomalies, handling missing values, and ensuring appropriate data transformation for time series analysis.

Part 3: Exploratory Data Analysis (EDA)

3.1 Visualization: Time series plot for each variables & Correlation Matrix

Key Findings

Our comprehensive time-series analysis reveals a competitive market where the brand's pricing strategies and competitors' moves intricately interplay. Here are the key findings with implications for strategic decisions:

  • Sales and Pricing Trends:

The brand's sales trajectory shows sensitivity to pricing fluctuations. Our data exhibit a clear pattern where RedStar's sales inversely correlate with its own price adjustments but interestingly improve with competitors' price hikes. This suggests a market perception of RedStar as a value proposition relative to competitors.

  • Advertising Influence:

A hiatus in RedStar's advertising efforts was observed, creating windows where competitors potentially capitalized on market attention. RedStar's advertising efforts have been found to impact competitors' strategies, suggesting an active market response to RedStar's marketing campaigns.

3.2 Descriptive Statistics

Key findings:

  • The summary table shows a statistically significant negative relationship between the brand's price and the brand's sales and a significant positive relationship between competitor's price and the brand's sales.

Part 4: Analyze the brand's performance and marketing effectiveness

4.1) Identify reactions of the brand’s advertising and price to competitors' advertising and price and vice versa.

This is done by analyzing which variables are temporally causing which other variables using 'Granger Causality tests.'

Granger Causality tests:

is used to find which variables are temporally causing which other variables.

The results show that:

1. The brand's price is Granger causing the brand's sales at lag 8

  • Changes in the brand's price have a predictive relationship with changes in the brand's sales that becomes apparent 8 periods later (8 weeks).

  • This suggests that it takes approximately two months for the full effect of price changes to reflect in sales numbers.

    2. The brand's advertising is Granger causing average competitor's advertising at lag 13

  • This suggests thatthe brand's advertising decisions influence or predict the advertising strategies of competitors, with this influence becoming evident after 13 periods (which could reflect a quarterly business cycle).

  • It could be that competitors are observing and responding to the brand's advertising initiatives with a time lag, indicating a reactive competitive environment.

    3. Competitor's price is Granger causing the brand's sales at lag 1

  • The competitor’s pricing decisions have an almost immediate effect on the brand’s sales (only 1 period lag).

  • This implies a very competitive market where consumers are quickly responsive to price changes from competitors, potentially due to easy comparison shopping or price sensitivity in the market.

    4. Competitor's price is Granger causing the brand's advertising at lag 13

  • This result means that competitor pricing strategies are able to predict fluctuations in the brand’s advertising expenditure with a 13-period lag.

  • The brand may be analyzing competitor pricing trends and strategically adjusting their advertising spend to counteract those pricing changes.

  • The delayed reaction that could align with strategic planning cycles or budgeting processes.

    5. Competitor's price is Granger causing the brand's price at lag 1

  • The brand's pricing strategy seems to be responsive to competitor pricing actions with a lag of only 1 period, indicating a potentially aggressive pricing strategy that quickly reacts to the competition.

  • This could reflect a highly competitive market.

  • These results suggest that both the brand and its competitors are closely intertwined in their strategic actions.

4.2) Identify business scenario based on the effects of 'own' and 'competitors' advertising and price actions on sales

This is done by analyzing on whether variables are stationary or evolving? We adopt

[1] Augmented Dickey-Fuller (ADF) Test (lag = 4)

[2] Phillips-Perron (PP) test

[3] KPSS Test

The results show that:

  • The brand's sales variable is "mean-stationary" (ADF test: p-value = 0.01 and PP test: p-value = 0.01 -> reject null hypothesis of unit root)

  • The brand's advertising variable is "mean-stationary" (ADF test: p-value = 0.01 and PP test: p-value = 0.01 -> reject null hypothesis of unit root)

  • The brand's price variable is "mean-stationary" (ADF test: p-value = 0.01 and PP test: p-value = 0.01 -> reject null hypothesis of unit root)

  • Competitor's advertising variable is "mean-stationary" (ADF test: p-value = 0.01 and PP test: p-value = 0.01 -> reject null hypothesis of unit root)

  • Competitor's price variable is "trend-stationary" (ADF test: p-value = 0.01 and PP test: p-value = 0.01 -> reject null hypothesis of unit root)

Key Findings:

The results imply "Business as usual" based on the four strategic scenarios for long-term marketing effectiveness of Dekimpe and Hanssens (1999).

  • Business-as-Usual combines stationary performance with stationary marketing.

  • In the case,the brand's sale (performance) is mean- stationary while other marketing mix of the brand and competitors are also stationary.

  • This long-term marketing scenario indicate that one-short marketing campaigns (such as a temporary increase in advertising spending or temporary price reduction) have only temporary effects on performance.

  • This results imply that the brand is competing in established market and brands, and therefore, marketing strategy efforts will only have temporary effect on the brand's sales.

4.3) Explore linear interdependencies of variables in the dynamic system.

  • The brand sales can be affected by its own advertising and price, as well as by competitors’ advertising and price. These variables, can all affect each other in a dynamic system. Moreover, sales can also be driven by exogenous effects of quarterly variation as well as a possible linear trend.

  • We adopt Vector Autoregression (VAR) Model to analyze analyzing multiple time series data.

    Key Findings:

    Current average comptitors' price:

    • The past average competitor price plays a significant role in determining its current value, and there is a positive trend in the average competitor price over time.

    The brand's advertising:

    • Even without the effect of the lagged variables, there's a constant effect on LnAdvertising.

    • This may suggests that the brand consistently invests in advertising or there's a consistent interest or awareness about the brand in the market.

    The brand's price:

    • There is a strong inherent downward trend or level in LnPrice, independent of the other factors in the model.

4.4) Identify significant dynamic impact of the different variables on each other? Focus on the significance of the effect variation as well as a possible linear trend.

We adopt 'IRF (impulse response function)' to analyze how one variable responds to a shock in another variable overtime.

Key Insights:

1. Reactive Market Dynamics:

  • The market seems to be highly reactive. The brand’s sales don't appear to be influenced by other factors within the observed confidence interval, which means they might have a stable demand or strong brand loyalty.

  • However, when it comes to pricing and advertising, The brand is sensitive to competitor actions, and vice versa.

    2. Pricing as a Competitive Tool:

  • Both the brand and its competitor seem to employ pricing as a competitive tool.

    3. Advertising to Counteract Pricing Movements:

  • The brand boosts its advertising when its competitor raises prices.

  • This suggests that the brand might use advertising to highlight its value proposition or specific promotions when it senses vulnerability in the competitor due to a price increase.

4.5) Find importance of each driver’s past in explaining observed variance in both the brand’s performance (sales) and marketing (advertising and price)le linear trend.

We adopt 'FEVD'

FEVD ( Forecast Error Variance Decomposition) is used to reveal how the forecast error variance in one variable can be explained by its own past shocks and all the shocks of the other endogenous variables. Therefore, it shows the relative importance of each variable in having contributed to the variation in the performance variable.

Key Insights:

1. Reactive Market Dynamics:

  • The market seems to be highly reactive. The brand’s sales don't appear to be influenced by other factors within the observed confidence interval, which means they might have a stable demand or strong brand loyalty.

  • However, when it comes to pricing and advertising, The brand is sensitive to competitor actions, and vice versa.

    2. Pricing as a Competitive Tool:

  • Both the brand and its competitor seem to employ pricing as a competitive tool.

    3. Advertising to Counteract Pricing Movements:

  • The brand boosts its advertising when its competitor raises prices.

  • This suggests that the brand might use advertising to highlight its value proposition or specific promotions when it senses vulnerability in the competitor due to a price increase.

Part 5: Conclusion: Managerial Insights & Recommendations

Managerial Insights:

Understanding the brand's Market Dynamics

Time- Series Analysis:

Our comprehensive time-series analysis reveals a competitive market where the brand's pricing strategies and competitors' moves intricately interplay.

1) Sales and Pricing Trends:

  • The brand's sales trajectory shows sensitivity to pricing fluctuations.

  • Our data exhibit a clear pattern where the brand's sales inversely correlate with its own price adjustments but interestingly improve with competitors' price hikes.

  • This suggests a market perception of the brand as a value proposition relative to competitors.

2) Advertising Influence:

  • The brand's advertising efforts was observed, creating windows where competitors potentially capitalized on market attention.

  • The brand's advertising efforts have been found to impact competitors' strategies, suggesting an active market response to brand's marketing campaigns.

Causality Test:

1) Price as a Leading Indicator:

  • The brand's pricing decisions at lag 8 emerge as a leading indicator for sales performance, affirming pricing strategy as a critical lever in forecasting sales outcomes.

2) Advertising Wars:

  • There's a two-way impact in advertising strategy - the brand influences competitors' advertising spend, and vice versa, especially at lag 13, indicating a potential market rhythm that savvy marketers can exploit.

3) Immediate Competitive Response:

  • Competitors' price strategies instantly affect the brand's sales and advertising response, highlighting the need for agility in the brand's marketing strategy.

Stationarity Test:

1) Stability in Change:

  • Despite market fluctuations, our tests reveal a "mean-stationary" status for all variables except for competitors' prices, which are "trend-stationary."

  • This underscores a stable competitive landscape where actions lead to predictable reactions within the market.

2) Temporary Shocks:

  • Given the stationarity of variables, we identify a "Business-as-Usual" scenario, where any marketing campaign yields only transient effects.

  • This implies that continuous innovation in marketing tactics is crucial to maintain market presence.

Dynamic Interactions and Long-Term Strategies

1) No Significant Long-Term Sales Impact:

  • Cumulative IRF analyses suggest that no variable has a lasting significant effect on the brand's sales, emphasizing the need for consistent and innovative marketing efforts to sustain market share.

2) The Ripple Effect in Advertising and Pricing:

  • Competitors' pricing changes positively influence the brand's advertising strategy over time.

  • The brand's own sales and advertising efforts, in turn, have a dynamic impact on pricing decisions, both for RedStar and competitors.

Variance in Performance and Marketing Drivers

1) Past as a Predictor:

  • The brand's sales and advertising activities are predominantly determined by their own historical performance, signaling a degree of predictability and the importance of understanding past trends for future strategy.

2) External Influence:

  • Though minimal, there's recognition that competitors' pricing strategies marginally influence the brand's advertising variance, suggesting that competitor price movements should not be ignored in strategizing marketing efforts.

Managerial Recomendations:

Strategic Pricing:

  • Leverage the brand's price elasticity by timing adjustments strategically in response to competitor price changes to optimize sales.

Advertising Agility:

  • Maintain a consistent advertising presence to prevent competitors from capitalizing on gaps. Monitor competitor activity to anticipate and swiftly respond to their advertising campaigns.

Innovation in Marketing:

  • Given the temporary impact of marketing strategies, the brand should prioritize continuous innovation to remain relevant and maintain market interest.

Data-Driven Decision Making:

  • Continue to refine marketing strategies based on historical data trends while staying attuned to immediate competitor moves.

Integrated Marketing Response:

  • Develop an integrated marketing strategy that encapsulates reactive agility and proactive planning, ensuring that RedStar remains a dominant voice in the conversation around value and quality.

By considering these insights, the brand can craft a more responsive strategy that not only reacts to the market but also anticipates changes, positioning the brand for sustained growth and profitability.