Historical Data Analysis of Business Strategies During Recessionary Times

I

t is no secret that lean Six Sigma (LSS) and design for Six Sigma (DFSS) have arguably been the most effective initiatives for improving bottom line results and revenue growth since the advent of Frederick Taylor and management science. Billions of dollars in bottom-line impact have been documented by those companies implementing it seriously^{1, 2}. The synergy between the two methodologies is also well known as many companies use the savings from operational efficiency derived from LSS to fund their longer-term product development efforts via DFSS¹.

During bad economic times, many quality gurus—along with substantial numbers of Champions, Green Belts, and Black Belts—have emphasized the need for either starting LSS and DFSS, or maintaining their impetus if already started. This philosophy, although recognized as needed by many leaders, has not been carried out by many companies during recessionary times. Instead, severe cost-cutting measures that focus on near-term results have tended to dominate the business landscape in recent months. What most advocates of the LSS-and-DFSS-initiatives-now-and-forever philosophy have yearned for is data to support their claims. Finally, such a longitudinal study has been conducted, and the data is quite revealing.

Undoubtedly a valid approach would be to study multiple economic recessions and see what the strategies of various companies were during those bad economic times and then correlate those strategies to business results following the recessions. That is precisely what researchers Ranjay Gulati, Nitin Nohria, and Franz Wohlgezogen did as reported in their article in the March issue of Harvard Business Review, “Roaring Out of Recession.” Their longitudinal study on recession strategies covered 4,700 public companies from Standard & Poor’s Compustat database. Thus, the sample size is considerable and provides legitimate support for this study. Large sample sizes are dear to the hearts of Six Sigma practitioners.

Furthermore, they studied the past three global recessions: 1980-1982, 1990-1991, and 2000-2002. Each company was studied three years prior to the recession, during the recession along with the strategies employed, and then three years after each recession. Each company’s post-recession performance is measured by sales (top-line growth) and profitability (bottom-line growth as measured by earnings before interest, taxes, depreciation, and amortization—EBITDA—as a percentage of sales). In statistical jargon, these would be considered response (or output) variables. The referenced article is structured such that one can glean three primary strategic variables (or input variables) that can be seen to affect performance.

Both the input and output variables considered in this analysis are depicted in the input-process-output (IPO) diagram below in figure 1. The purpose of this article is to provide a more detailed analysis of the relationships between the input and output variables using a quasi-designed experiment (i.e., coded historical data) approach.

Figure 1: Input-process-output diagram

Analysis

It’s well known that LSS is a primary driver for operational efficiency (input variable A) and that DFSS is a primary driver for new product and asset development (input variable C)^2,4,5. Thus, these two initiatives are directly related to two of the three inputs shown above in figure 1. While it could be argued that LSS is also associated with head-count reduction (variable B), that is not the direct intent of LSS, though it can be a natural consequence of LSS. The referenced article treats variable B as a major cost-reduction factor but does not allude to any part of it as being a result of operational efficiency. Hence, this analysis will treat the two as separate variables, although their individual effects cannot be completely separated as will be shown later. The authors of the article broke each of the input variables into three possible values according to the bottom, middle, and upper thirds of the population, with all values being normalized based on the revenue and size of the companies. Interestingly, whether intentional or not, this clustering technique essentially codes the variables into three values (-, 0, +) as specifically defined below, lending itself to further analysis of the data via coded historical data.

 

A: Operational efficiency

(-) Invested much less than the middle third tier of companies in operational efficiency (i.e., are in the bottom third)
(0) Is in the middle third of companies investing in this category
(+) Invested much more than the middle third tier of companies (i.e., are in the top third)

B: Head-count reduction

(-) Reduced head count much less than the middle tier of companies in this category
(0) Is in the middle third of companies with regard to head-count reduction
(+) Reduced head count much more than the middle third tier of companies in this category

C: New product and asset development

(-) Invested much less than the middle third tier of companies in this category
(0) Is in the middle third of companies investing in this category
(+) Invested much more than the middle third tier of companies in this category

 

The authors in the Harvard Business Review article categorized the business strategies implemented during recessionary times as follows. These definitions link the business strategies to the variables and their settings.

Prevention-focused companies are very defensive minded and primarily use cost-reduction methods as their strategy. From the variable point of view, this correlates to heavy on operational efficiency and head-count reduction but no more than average on new product and asset development.

That is, A=+, B=+, and C=0.

Promotion-focused companies are those who invest heavily in the offensive strategy, namely investing in new product and asset development, but do no more than average in the cost-reduction methods.

From a variable setting point of view, this equates to A=0, B=0, and C=+.

Pragmatic companies are those who use a combined offensive and defensive strategy.

That is, from a variable point of view, these companies are in the upper third of each of the variables, namely A=+, B=+, and C=+.

Progressive companies are those that have attempted to optimize across both the offensive and defensive strategies.

That is, these are companies that invest heavily in operational efficiency (A=+) and in new product and asset development (C=+) but who have only been in the middle third or average in head-count reduction (B=0).

The effects of the various strategies are shown in figure 2, taken from the aforementioned article. The sales and EBITDA percentages are the compounded annual growth rates for the three-year period after the recession.

Figure 2: Compounded annual percentage growth rate for three years after the recesssion³

 

A more detailed analysis can be conducted if each of the recession strategies is defined in terms of the input variables shown previously in the IPO diagram. Similar to a design matrix, we summarize each of the strategies in the figure 3 below by way of their coded settings as defined above. Note that negative (-) settings are not used. A substantial percentage of those companies didn’t survive, so the coded settings reflect only nominal (0 or middle third) companies and aggressive (+ or top third) companies in each of the variable categories of A, B, and C. The response variables of sales and EBITDA are also shown with their respective variable combinations.

		Variable Settings
Strategy Name	Focus	A	B	C	Sales		EBITDA
Nominal or Average	Companies in general	0	0	0	6.6		6.2
Prevention	Cost reduction only	+	+	0	6.3		4.4
Promotion	New products only	0	0	+	7.9		6.2
Pragmatic	Balanced approach	+	+	+	9.4		9
Progressive	Optimized balanced approach	+	0	+	13		12.2

Figure 3: Summary of strategies

By looking at the “design matrix,” or the columns labeled A, B, and C, in figure 3, we see some interesting characteristics. The first four runs (or rows) of this matrix show basically a full-factorial design for the A and C columns as well as for the B and C columns. A and B are aliased (or totally confounded) in the first four runs, and this will prevent us from evaluating their effects independently. Fortunately, row 5 allows some decoupling of the effects of A and B when A=+/C=+ and allows us to observe some important effects. The graphical analysis that follows uses all five runs in the matrix.

The marginal means plots in figure 4 below show that factor B (head-count reduction) has a different effect than do factors A and C. While aggressiveness in variables A and C seemed to improve sales and EBITDA, aggressiveness in B seemed to have the opposite effect, namely reducing sales and EBITDA. While some head-count reduction (nominal or B=0) may be needed during recessionary times, being overly aggressive (B=+ or B=1) in this variable seems to have a longer-term negative effect. This may be due to the time it takes to rebuild the necessary infrastructure, including reinvestment in intellectual capital and the ill-effects of disillusionment throughout the remaining work force after a substantial downsizing. It is also interesting to note that although the slopes of the sales and EBITDA lines are relatively the same, EBITDA lags sales in absolute percentages for all three variables. This apparent inefficiency in converting sales into profitability, which may be related to the phenomenon just mentioned, indicates that any reduction in operating costs attained during the recession weren’t sustained on a permanent basis. The relationship between sales and EBITDA can also be seen in the bar chart of the different strategies.

 

Figure 4: Marginal means plots for A, B, and C

 

The interaction plots in figure 5 below provide a glimpse of A and C having a greater interaction effect than B and C, on both sales and EBITDA. The bump that aggressiveness in A and C have on sales and EBITDA is apparent, while aggressiveness in B and C don’t provide that same kind of positive influence on the response variables. In fact, the response is somewhat negative when B and C are aggressively pursued, potentially because the necessary resources aren’t available to carry out the aggressive approach to new product/asset development.

Figure 5: Interaction plots

 

It is possible to isolate the effect of B in the case when A and C are aggressively pursued, that is, when A=1 and C=1. This is simply the difference between runs 4 and 5 in the table and is graphically in figure 6 below.

Figure 6: Marginal means for B when A=1 and C=1

 

These graphical results show that a lean Six Sigma approach that positively enhances operational efficiency (factor A) and implementing design for Six Sigma, which should have a direct positive influence on factor C (new product and asset development), are practical strategies that should not be delayed during recessionary times. It also shows that these two strategies can have a dampening or mitigating effect on the negative impact of a very aggressive downsizing strategy (factor B). Thus, it behooves leaders to understand that LSS and DFSS provide practical long-term benefits when implemented properly in good and bad economic times. This longitudinal study seems to validate what Deming said many years ago in the first of his 14 Points for Management: “Create a constancy of purpose toward the improvement of product and service. Consistently aim to improve the design of your products and services. Innovation, money spent on research and education, and maintenance of equipment will pay off in the long run⁶.” It is unfortunate that many companies haven’t heeded Deming’s points and have instead pursued strategies that don’t result in longer-term optimal results.

To summarize, this paper uses a clustering technique to classify business strategies in terms of combinations of variable settings. It has associated LSS with operational efficiency (factor A) and DFSS with new product/asset development (factor C). Although companies considered in this study may have employed initiatives other than or in addition to LSS and DFSS to enhance these strategies, it is an established fact that LSS and DFSS are successful drivers for improved operational efficiency and new product/asset development.^1,2,4,5 Hence, this analysis provides substantial evidence that LSS and DFSS are useful strategies to be employed even in down times.

Leaders within industry and government sectors should understand and necessarily consider the long-term consequences of making across-the-board cuts and paying little attention to initiatives that could in fact be the key to post-recession growth. There is no longer any excuse for not getting a handle on cause-and-effect relationships within a business environment.

Finally, this article illustrates that tools commonly used by LSS and DFSS practitioners at a tactical level can also be used to efficiently collect and analyze data in a more strategic setting.

 

References

1. “The Next Big Thing,” by Roger Hoerl and Ronald Snee (Six Sigma Forum Magazine,
February 2010)

2. “Six Sigma Saves a Fortune,” by Michael Marx (iSixSigma Magazine, January/February 2007)

3. “Roaring Out of Recession,” by Ranjay Gulati, Nitin Nohria, and Franz Wohlgezogen (Harvard Business Review, March 2010)

4. “Duplicating Success,” by Quincy Allen and Len Parker (iSixSigma Magazine, September/October 2006)

5. “Using Design for Six Sigma,” by Jonathan Atwood (iSixSigma Magazine, July/August 2005)

6. Knowledge Based Management, Second Edition, by Mark J. Kiemele, Richard C. Murrow, and Lee R. Pollock (Air Academy Associates: 2007); paraphrased from Out of the Crisis, by W. Edwards Deming (MIT, 2000)

 

About The Author