Five Weekends a Month Matters: The Case of New Automobile Sales in México

In Mexico, monthly sales of new cars improve when the month has five weekends. This can be due to the fact that families buy weekends. Although the sale is recorded the following month, by administrative factors. Here the statistical validation of that calendar effect.

Basics

In theory, the degree of dynamism of domestic car sales, in a particular market, is explained by the following factors:

• Available income and patrimonial situation of families
• Degree of access to credit
• Seasonality
• Sales prices and degree of competence
• Consumer confidence
• Levels of insecurity

In addition, in more detail, other factors can also play a role. Such as weather conditions, promotions and social situation at the time of purchase, and so on. Identifying these additional factors and validating them helps to interpret the sales indicator better, with as little noise as possible, and can even help to make better sales forecasts.

Here the effect on the sales of the months with five weekends was valid. For the moment, when observing the following graph we can find this calendar effect:

 

Sale of New Cars

(Percentage Annual Change)

Souce: INEGI.

As I mentioned before, during the analysis period there was an improvement in sales in the month after the one with five weekends. Only this regularity breaks significantly August. 2017. There, it is probable that the effect of the negative trend dominated the positive effect of the calendar effect described here.

Derived from the above, I statistically test the role of this variable. Which I describe more extensively in the statistical addendum. From the results of the model we can say the following:

• Months with five weekends have a positive impact on sales. In particular, the following month, when increasing them around 5 percent at annual rate.
• The incorporation of this information can help to interpret the situation of the sector and could even help in fine-tuning sales forecasts.
• There are several areas of opportunity in other areas, such as sales of other products or other economic activities, to consider this kind of calendar effects.
• In addition, there are other variables to explore, in the case of car sales, such as weather.

So, to try to validate this factor, I estimated, via econometric tools, an equation to discover the net impact of the five weekends on car sales. In the following section.

Statistical Section

Proposed equation

Next the mathematical function of the sales of cars used for the validation of the hypothesis:

Sales = Rest of Factors + Months with 5 Weekends + Stochastic Error

In the econometric model in the intercept and the lagged sales variable captures the rest of the factors not related to the number of weekends. And I also incorporate a dummy or dichotomous variable for the months that enjoy five weekends.

In precise comment that although the model has unitary root, I expose the model without differences for expository purposes, if you want the results in the case of the differentiated model I can send them, my email is salva.rguez@gmail.com. These results, of the reliable model, are congruent with the one presented here.

Results of the Econometric Model

The estimated model indicates that the months with five weekends have a positive and statistically significant effect on the coming month. In particular, it increases sales by 5.07 percent the following month at an annual rate.

Next the estimated equation, the results in numerical terms and in the graph that shows the degree of adjustment of the statistical model with respect to the information -series in annual variations-: in red the original data, in green the result of the model and in blue the forecast error. The period contemplated goes from January 2010 to August 2017, where 31 months with five weekends occurred, and are national data adjusted for seasonality.

Estimated Model:

Sales = 0.729 * Trend (-1) + 5.07 * Months with Five Ends (-1) + 1.617

Statistical results: