Archive for the ‘Risk Analyses’ Category

Measuring Risk in Collective Bargaining

Monday, May 31st, 2010

There are few business processes as fraught with risk as collective bargaining.  Although rare, strikes and lockouts are always possible outcomes of labor negotiations.  For most organizations engaged in collective bargaining, a strike or lockout usually results in some part, if not all, of a business operation shutting down either temporarily or permanently.  Excluding acts of God, war, or government-mandated actions, no other business process carries such high risks to continuing operations than labor negotiations.

Yet, given this reality, businesses and organization rarely analyze their negotiations from the perspective of risk, except on an ad hoc, spur-of-the-moment, “from the gut” determination when considering and developing proposals, usually at the at the last minute in negotiations.  Often, this leads to either supremely risk-adverse tactics that result in expedient over-paying in the final settlement, or, at the opposite extreme, very risky bargaining tactics that expose organizations to the very real possibility of operational shut down, usually accompanied with relatively small marginal bargaining returns.

How can labor negotiators properly incorporate risk measurement into their bargaining analyses, tactics, and strategies?

LCA has developed an analytical approach that quantifies bargaining risk within the negotiations costing process.  Thus, risk is treated as either a premium or an offset to the overall economic cost of a proposal, based on the projected probability of a work stoppage resulting from said proposal.  This analytical process produces a “risk-adjusted decision cost” (“RADC”) that when compared to RADCs of other proposals, allows a negotiator to judge proposals independent of risk (i.e., on an “apples-to-apples” basis with the complicating factor of risk removed).

By using RADCs, a negotiator can objectively ascertain whether pushing a seemingly more risky bargaining approach is worth the associated potential bargaining returns.  Or to put it in more realistic terms by way of an example, is trying to reduce the amount of a wage increase by, let’s say, a dime (10¢) in order to save some determinable amount of aggregate wage dollars, worth the fact that pursuing this savings may significantly increase the chance of a strike, leading to substantial loss of sales and extra strike costs unexpectedly incurred by the organization.  Only with an understanding of the relationship of costs/savings and risk can a negotiator properly make this decision.

So how do we measure the risk associated with collective bargaining?  LCA has developed a formula comprised of three components:  the underlying cost or savings of a proposal unadjusted for risk (i.e., a “normal” labor costing) multiplied by the product of the projected probability of a work stoppage [“Pr(strike)”] and the projected potential cost of a work stoppage (i.e., loss of revenues, extra strike costs, etc.).  Put into mathematical terms, the formula is:

Negotiations Cost/(Savings)  ×  Pr(strike)  ×  Projected Strike Costs.

The “Negotiations Cost/(Savings)” is an objective number provided in a standard labor costing.  “Projected Strike Costs” are also fairly objective, and are usually determined by the financial accounting function of an organization as part of strike contingency planning.  The projected probability of a strike, “Pr(strike)”, is more subjective and requires intimate knowledge of the labor negotiations and of the union(s) involved, preferably over more than one contract period.  LCA can develop a strike probability distribution by interviewing an organization’s key personnel responsible for decision-making in collective bargaining.  This interview process entails asking numerous “what if” questions of the key bargaining personnel:  for example, what if management offers $X over three years, what is the chance of a strike?; what if the offer is $X+?; etc.  By relating potential economic outcomes of major negotiating issues (e.g., wages, fringe benefits, operational provisions, etc.) to the perceived chance of a strike, LCA will construct such a distribution of the probability of a strike versus the value of potential settlements.  With these three components determined, each potential settlement proposal can be vetted not only for economic impact, but also, also for potential risk to the operation by comparing its RADC to those of others.  Such comparisons can illuminate the trade-offs between marginal costs/(savings) and overall potential risk.

As an example, I have developed the following strike probability distribution using my extensive experience in the retail food industry.  To keep it simple, I have limited the bargaining outcomes to just wage increases (although, in reality, there are many additional, crucial economic issues, foremost among them, health care benefits and costs).  I first set up a graph tracking my predictions for a successful bargaining settlement versus the total wage increases over the contract term (note that in this industry, wage increase usually are settled at between 30¢-40¢ per year, for a three-year total of 90¢-$1.20).  Below is the resulting graph depicting the probability of settlement [“Pr(SETT)”].



Next, just for the sake of understanding and confirming my settlement predictions, I graphed my estimate of the probability for rank-and-file ratification of the settlement [“Pr(RAT)”].



Note that Pr(RAT) roughly tracks Pr(SETT), but with some significant deviations reflecting the reality that union leadership can and will often use their political influence in ratification votes, increasing the chances of economically favorable settlements and decreasing the chances of settlements perceived as less favorable.  In fact, the areas between the Pr(RAT) and Pr(SETT) curves can be construed as influence areas, with a “positive spin zone” occurring when Pr(RAT) is greater than Pr(SETT) in the higher dollar value part of the graph, and, similarly, a “negative spin zone” occurring when Pr(RAT) is less than Pr(SETT) in the lower dollar value part of the graph, as shown below:



Now to Pr(strike).  Obviously, the probability of a strike is indirectly proportional to both Pr(SETT) and Pr(RAT), but on what basis.  In the early days of labor relations, when strikes were more commonplace and more effective, it could be assumed that a strike would be called whenever a settlement was not reached.  In strictly mathematical terms, this could be expressed as a pure complementary equation as follows:

Pr(SETT)  +  Pr(theoretical strike)  =  100%

Simply put, the above equation means that there would be an 100% chance that the result of collective bargaining would be either a settlement or a strike (I used the term “theoretical strike” so as to avoid confusion later with the actual probability of a strike used in my example).  Put another way, the equation would read:

Pr(Theoretical Strike)  =  100%  –  Pr(SETT)

Graphed against our earlier Pr(SETT) and Pr(RAT) curves, this Pr(theoretical strike) curve would look like the graph below:



However, in reality, strikes are now rare due to their repeated ineffectiveness.  Just as we now know in the world of foreign policy after the Cold War that the probability of nuclear war is considerably less than the probability of conventional war, similarly, the threat of a labor strike is much less than other forms of labor protest (e.g., corporate campaigns, work-to-rule actions, etc.).  Indeed, strikes are often referred to as “going nuclear” by both sides of the bargaining table, due to the fact that their results, either positive or negative, are always irreversible.  Considering this, our Pr(strike) curve must be substantially depressed from the Pr(Theoretical Strike) curve set forth above.  In my retail food example, I would set it as follows:



By either determining the algebraic function of Pr(strike) or constructing a table of approximate probability value for it, one can come up with the projected probability of a strike for all economic outcomes.  This is a key requirement in using our RADC formula to properly incorporate risk in analyzing negotiations options.

The final step in our risk analysis process is determining the projected cost of a strike.  Work stoppages are indeed expensive because they not only incur additional operational costs, but usually entail a significant loss of revenues as well.  Yet, companies are sometimes willing to take a strike because it makes financial sense given the economic issues being determined in labor negotiations.  Again, only by calculating the projected costs of a strike can such a determination be made.  The costs of a strike vary by organization and industry, but can include the following: 

  • Loss sales and revenues;
  • Labor costs of replacement workers;
  • Logistical costs of contingent, non-bargaining unit employees doing bargaining unit work;
  • Logistical costs of replacement workers; and,
  • Legal and administrative costs of operating during picketing and implementing a strike contingency plan.

In addition, there are some cost savings during a strike, which should be netted out against other additional strike costs, primarily the saved wages and fringe benefits of striking employees.  Cost accountants can usually project these additional costs and savings of a potential strike, as well as the expected loss in revenues during a strike.  The only challenging projection is the length of the strike itself, but by using previous strikes within an industry as a guide, a realistic estimate can be made.

With all of our three components determined, various RADCs can be calculated.  They can be compared to one another to determine which proposals offer the best economic outcome at an acceptable risk.  By dividing the RADC from the unadjusted-for-risk cost, we can also come up with a risk factor for the proposal, with the obvious implication of the higher the risk factor, the larger the risk involved in attaining such a proposal.

To view an example of the analysis described above, please see our page on “Risk Analyses.”




Next month:  “The (In)Finite Power of Lump Sums”