University of Limerick, 2002.
“The Relationship Between Concentration, Turnover and Capital Expenditure”
Name: Emily Flynn 9924663
Siobhán Purcell 9923039
Karen O’Sullivan 9922415
Module: Industrial Economics
Lecturer: Dr. Bernadette Andreosso-O’Callahan
Table of Contents
1. The Variables.
1.1 The Market
1.2 Concentration
1.3 Turnover
1.4 Capital Expenditure
2. Computer Commands Used.
2.1 Data Entry
2.2 Defining the Y-Intercept
2.3 Running the Regression
2.4 Testing for Correlation and Scatter Plots
3. Results from the Regression.
3.1 Statement of the Regression
3.2 Relationship Analysis
4. Interpretation.
4.1 Theoretical Support
4.2 Criticisms and Limitations
4.3 Conclusion
5. Appendices
The Variables.
When choosing the definitions for our variables, we decided to base them upon the works of J.S. Bain, one of the pioneers of the Structure – Conduct – Performance paradigm. His first study, in 1951, found that as concentration increased, profits increased. The second study, in 1956, found that profits should be higher in industries with high concentration and high barriers to entry.
1.1 – The Market.
We defined our market as being all companies listed on the Hoovers website[1] (as of 12th December, 2002), for a particular industry where financials were available for 2001. In both of Bain’s studies, he used cross sectional data for a given year and this is why we took a similar path. Our primary market was in the United States. Some of the multinational companies, however, only produced consolidated financial statements and we were unable to extract data that was US specific.
1.2 – Concentration.
Concentration is a measure of the intensity of the measure of competition or control[2]. In our study, we used the four-firm concentration ratio (CR4). This is defined as the cumulated market share of the four leading firms in the industry[3].
1.3 – Turnover.
Industry turnover was found to be the total sales for all companies in the industry (where financials were available) for the year 2001, in US dollars. CR4 turnover, was the cumulative total sales for the four leading firms in the industry.
1.4 – Capital Expenditure.
This is defined as money spent to acquire or upgrade physical assets such as buildings and machinery[4]. In cases where this data was unavailable, Cash Flows from Investment Activity was used as a proxy. Shepherd (1990) identifies capital requirements as an exogenous barrier to entry.
Computer Commands Used.
2.1 – Data Entry.
In order to perform the regression, we used Microfit™ 4.0. Firstly, we manually entered the data into Excel, and then pasted into Microfit (see appendix XXX). We chose the “Undated” option from the Data Frequency window. We entered “29” into the number of observations text box, and “3” into the number of variables text box. In the additional clipboard information, we chose the “Variable names with optional descriptions on the first line, then data” option and also the “Dates in the first column, then data” option.
2.2 – Defining the Y-Intercept.
Next, we defined the Y-intercept. To do this, select “Process” and then select “Constant”. We called the Y-intercept “INPT”.
2.3 – Running the Regression.
Following this we ran the regression by clicking on “single”. We were then prompted to list the dependent variable followed by the independent variables and the intercept. To run the regression, we clicked on “start”. We ran three regressions to determine which regression best fitted the theory (as detailed in section four).
The three regressions are as follows:
TURNOVER INPT CAPEX CONCENTRATION
CAPEX INPT INPT TURNOVER CONCENTRATON
CONCENTRATION INPT TURNOVER CAPEX
(see appendix XXX for the results.)
2.4 – Testing for Correlation and Scatter Plots.
Next we tested the variables for correlation. The command for this is “COR Variable 1 Variable 2”. For example “COR CAPEX CONCENTRATION”.
Lastly, we scatter plotted the variables against each other. The command for this is “SCATTER Variable 1 Variable 2”. For example “SCATTER CAPEX CONCENRATION”. (See appendix XXX)
Results from the Regression.
3.1 – Statement of the Regression.
Turnover = 284724.9 + 35.7615Capex – 437445.3Concentration.
S.E. (75447.3) (5.3654) (106164.0)
3.2 – Relationship Analysis.
Intercept: If no units were sold, turnover would be (on average) 284724.9 dollars, but this situation is unlikely to occur in practice.
Capital Expenditure: As turnover increases by one unit, capital expenditure increases on average by 35.7615 dollars.
Concentration: As turnover increases by one unit, concentration decreases by 437445.3 per cent.(This is an obvious error of the regression).
R2 = 0.6878. This implies that 68 per cent of the dependent variable (turnover) is explained by the two independent variables (capital expenditure and concentration). While this result is does explain a sizeable proportion of the turnover, the fact that it does not explain 32per cent indicates, again, that variables are missing.(See appendix XXX).
T-Statistics: It was found that the t-statistic (for both turnover and capital expenditure) was greater than the t-critical value at both the 5 per cent and 1 per cent levels, thus making it statistically significant.
Scatter Plots: there are obvious outliers in all of the scatter plots (see appendix XXX).
Capital Expenditure on Concentration: from this scatter plot there is no obvious linear relationship between the two variables.
Turnover on Concentration: from the first scatter plot, there appeared to be a slight linear relationship, but with two distinct outliers. When these outliers were removed from the data, however, there was no apparent relationship between the two variables. This indicates that the relationship could be industry specific.
Turnover on Capital Expenditure: there are five observable outliers in this plot. When
these are removed, the remaining industries exhibit a linear relationship.
Interpretation.
4.1 – Theoretical Support.
As previously mentioned, Bain’s work was at the forefront of research in this area. Mann (1966)[5], while building on many of Bain’s original assumptions, also investigated the relationship between profit and barriers to entry. He found that industries with “very high” barriers to entry enjoyed higher profits than those with lower barriers to entry. (Weiss, 1974)[6] found that there was a considerable relationship between concentration, barriers to entry and profit.
Entry barriers are one of the key methods of explaining profitability differences. Concentration can create entry barriers; high levels of concentration may be conducive to high prices and high levels of profits. Unless there are some appreciable barriers to entry, high profits will naturally attract new entrants who will eventually undermine pricing co-ordination. Monopoly profits may be realised in an industry with low concentration, if there are significant barriers there to restrict entry. The Major Drugs Manufacturers are a prime example of this type of activity, with most companies having high levels of capital expenditure and low concentration. Asplund (2000) also found that capital expenditure and sunk costs are often emphasised as a determinant of entry/exit decisions and strategic investment[7].
Output measures of concentration are correlated with the error term because of the combination of two factors. First, demand, factor prices, and other “basic conditions” directly affect both price and measured concentration. Second, these basic conditions either are omitted variables or are observed with error. The combination of these two forces causes measured concentration to be correlated with the error term in the estimating equation, and that causes the OLS estimator of the effect of concentration to be biased. The OLS estimator reflects not only the effect of structure but also the effects of the unobserved components of the basic conditions[8]. Asplund and Sandin, however, found that the negative relationship between market size and concentration can be expected to hold for markets with exogenous sunk cost (i.e. capital expenditure)[9].
4.2 – Criticisms and Limitations.
The industry was defined as those companies listed on the Hoovers website. In reality this is not completely applicable, as some smaller and private companies are excluded. This is an indication of information asymmetry.
The intrinsic nature of the different industries was not taken into account. This could explain the random outliers, for example major drugs manufacturers. In this industry, the concentration was very low at 0.158 but the turnover was one of the highest surveyed. Capital expenditure was also high across the board. In contrast to this, the computer hardware industry displayed a high concentration of 0.923, high turnover and high capital expenditure levels. Petroleum products had low concentration of 0.327 combined with low levels of turnover and capital expenditure. McGahan and Porter (1999) found that for a large sample of US industries during the time from 1981 to 1994 that industry-specific effects dominate firm specific effects[10].
In Bain’s work, he used the CR8 index to measure concentration in an industry. In this study, the CR4 index was used. Bain focused on whether firms operating in industries with CR8 greater than 70 corresponding approximately to CR4 above 50 were more profitable than firms operating in lower concentration industries[11].
We found a negative relationship between concentration and turnover, and this deviates from Bain’s work and also that of Mann (1966) and Weiss (1974). Bain also used profitability as one of his variables but in our study we used turnover in its place.
Some companies observed produced multiple products across industries, and this meant that we were unable to discern a single market (e.g. Phillip Morris U.S.A. in the tobacco industry also owns 84 per cent of Kraft Foods). This horizontal diversification exaggerates their turnover and as a result, the concentration within a given industry.
This is a basic relationship, only taking into account three variables. It omits a number of variables which may have a substantial effect on the outcome of the regression (R2). These include geographic dispersion, buyer concentration, R&D, price elasticity of demand and advertising.
We have assumed a linear relationship, perhaps indices should have been used; the relationship may be curve-linear. Two of our variables, turnover and concentration exhibit a negative correlation. This indicates that a dummy variable should be introduced into the relationship. Turnover and capital expenditure exhibit a high positive relationship, also indicating again that there is a variable missing from the equation.
The data in this regression was only taken from 2001. Due to this limitation we are only able to discern the magnitude of this relationship for one year, and we do not know if it endures over time. Brozen (1971) criticised Bain’s work in this respect. He found that “for the 42 industries of Bain’s initial 1951 study, the profit difference of 4.3 per cent found between highly concentrated and less concentrated groups diminished to only 1.1 per cent by the mid fifties”.
4.3 – Conclusion.
Theory supports the supposition of a relationship between concentration, turnover and capital expenditure. This study found that there was a significant relationship between the variables, with a high R2.This high R2, however, did not explain approximately 30 per cent of turnover. This leads to the conclusion that our results were spurious in nature and signifies that there is an omitted variable or that our variables are observed with error.
Bibliography:
Books:
Douma, Sytse and Schreuder, Hein, Economic Approaches to Organisations (2nd Ed.), 1998, Prentice Hall, Great Britain.
Griffiths, William, E., Hill, R. Carter and Judge, George G, Learning and Practicing Econometrics, 1993, John Wiley and Sons, New York.
Gujarati, Damodar, Essentials of Econometrics, 1992, McGraw-Hill Inc, Singapore.
Jacobson, D and Andreosso-O’Callahan, B, Industrial Economics and Organisation, A European -Perspective, 1996, McGraw-Hill, London.
Neumann, Manfred, Competition Policy, History, Theory and Practice, 2001, Edward Eglar Publishing Ltd., UK.
Perloff, J.M., and van ‘t Veld, Klaas, Modern Industrial Organisation (2nd Ed..), 1994, Harper Collins, New York.
Scherer, F.M. and Ross, D, Industrial Market Structure and Economic Performance (3rd Ed.), 1990, Houghton Mifflin Co., Boston.
Schmalensee, Richard, and Willig, Robert (Ed.’s), Handbook of Industrial Organisation Volume 1, 1989, Elsevier Science Publishers, Amsterdam.
Journals:
Asplund, M. and Sandin, R., The Number of Firms and Production Capacity in Relation to Market Size, The Journal of Industrial Economics, VOL. XLVII, March 1999, page 69-85
Asplund, M., What Fraction of a Capital Investment is Sunk Costs?, The Journal of Industrial Economics, Vol. XLVIII, No. 3, September 2000, page 287-304
Evans, W.N. et al, Endogeneity in the Concentration-Price Relationship, The Journal of Industrial Economics, Vol. XLI, No. 4, December 1993, page 431-437.
McGahan, A and Porter, M.E., The Persistence of Shocks to Profitability, Review of Economics and Statistics, 1999, 81, 143-53.
Websites:
www.hoovers.com
www.sec.gov
www.investerword.com/cgi-bin/getword.cgi?703
9
[1] www.hoovers.com
[2] Jacobson, D. & Andreosso-O’Callahan, B, 1996, page 53.
[3] Ibid, page 54.
[4] www.investerword.com/cgi-bin/getword.cgi?703
[5] Perloff, J.M., and van ‘t Veld, Klaas, Modern Industrial Organisation (2nd Ed..), 1994, page 350
[6] Ibid, page 351.
[7] Asplund, M., What Fraction of a Capital Investment is Sunk Costs?, The Journal of Industrial Economics, Vol. XLVIII, No. 3, September 2000, page 287-304
[8] Evans, W.N. et al, Endogeneity in the Concentration-Price Relationship, The Journal of Industrial Economics, Vol. XLI, No. 4, December 1993, page 431-437.
[9] Asplund, M. and Sandin, R., The Numberof Firms and Production Capacity in Relation to Market Size, The Journal of Industrial Economics, VOL. XLVII, March 1999, page 69-85
[10] McGahan, A and Porter, M.E., 1999, The Persistence of Shocks to Profitability, Review of Economics and Statistics, 81, 143-53.
[11] Scherer, F.M. and Ross, D, 1990, Industrial Market Structure and Economic Performance (3rd Ed.), Houghton Mifflin Co., Boston.