Using Regression Analysis in Evaluating Catalyst Markets

An extensive research of the Internet will find many estimates of annual global market sales values for catalysts.  For example, the following table 1 shows values that I found on the Internet:

year	sales in   $ billion
2001	10.5
2001	10.2
2005	11
2006	14
2007	13.5
2009	16.5
2009	14.89
2010	17.5
2010	14.04
2011	21
2012	19.2
2014	16.3
2015	16
2015	20.41
2015	19.2
2015	20.41
2016	19.5
2016	19.5
2017	26.4
2018	24.1
2018	24.1
2018	20.6
2019	21
2020	27.59
table 1

These estimates are made by market research firms, who then post them on the Internet.  Estimates are also found in technical articles on various aspects of catalyst use.  I have no idea how these estimates are made, but I assume that good approaches and reasonable analysis are used.     However, you can see in table 1 that variations exist in the estimates.  For example, one estimate for 2010 is $17.5 billion and one is for $14.04 billion.  I did a regression analysis on the data in table 1 and found that the variations in the data are reasonably small.  For example, the R-squared value is 0.78.  This tells me that although the estimates vary, the variations are small and support the conclusion that the various determinations are consistent with one another.

The regression analysis also, by providing coefficients a and intercepts b for the linear equation y = ax + b, allows for computing market sales value y for each year x.  In addition, the likelihood, at a 95% probability, that the market sales amount will be in a range of y values can be determined.

The following table 2 shows the y values for years from 2001 to 2020, based on the linear equation, y = ax + b. 

year	sales in billions
	y - std. dev.	y	y + std. dev.
2001	7.3	9.5	11.8
2001	7.3	9.5	11.8
2005	10.3	12.6	14.8
2006	11.1	13.3	15.6
2007	11.9	14.1	16.4
2009	13.4	15.6	17.9
2009	13.4	15.6	17.9
2010	14.2	16.4	18.7
2010	14.2	16.4	18.7
2011	14.9	17.2	19.4
2012	15.7	17.9	20.2
2014	17.2	19.5	21.7
2015	18.0	20.2	22.5
2015	18.0	20.2	22.5
2015	18.0	20.2	22.5
2015	18.0	20.2	22.5
2016	18.8	21.0	23.2
2016	18.8	21.0	23.2
2017	19.5	21.8	24.0
2018	20.3	22.5	24.8
2018	20.3	22.5	24.8
2018	20.3	22.5	24.8
2019	21.1	23.3	25.5
2020	21.8	24.1	26.3
table 2

Also shown is the range of y values that one can expect the correct value to be within, with a 95% likelihood, based on the data in table 1.   For example, for 2010, the overall market value for all catalyst sales likely fell (with a 95% confidence level) within a range of $14.2 billion to $18.7 billion.

The total market value for all catalyst sales are often broken down into 4 categories of catalysts: catalysts for refineries; catalysts used in polymerization; catalysts used in chemical synthesis; and catalysts used in emissions control.   And estimates of annual sales values for these 4 categories, like the estimates for the overall sales value for all catalysts, can be found on the Internet.  

I did a regression analysis on each of the sets of data for these categories, out of which I was able to create the following graphs:

You can see from the information in these graphs that the R-squared values are reasonably good.  Also, very interesting, is the x coefficient (a in the equation y = ax + b) values for each category.  The x coefficient is an indication of the slope of the line; the expected rate of increase of the y value (annual sales amount).  In other  words, the x coefficient value can be viewed as an indication of the expected sales increase with time – the higher the coefficient, the greater the expected annual increase in sales for the category.  From the graphs, x coefficients values determined are: emissions control, 0.60; refining, 0.21; chemical syntheses, 0.21, and polymerization, 0.11.

That emissions control shows a much higher expected rate of annual sales increases is consistent with what is reported on the Internet that emissions control is the fastest growing catalyst group.