HMT Framework for Labor Markets
Revisiting basic principles
On the product market side, market definition is fundamentally about substitution between products. The goal in defining markets is to separate close substitutes (products to include in the market) from more distant substitutes (products that can be left out of the market). The closeness of substitution between products is defined based on consumer behavior: i.e., how readily consumers would switch between the products if their relative prices changed. An appropriate market includes a set of substitutes that are close enough to provide competitive constraints on the conduct at issue.
The HMT is a commonly used framework that formalizes this exercise of identifying close substitutes. The HMT asks whether a hypothetical, profit-maximizing monopolist in the proposed product market would increase prices by at least a small but significant amount over a sustained (non-transitory) period.
The profitability of a price increase for the hypothetical monopolist depends on substitution: what share of unit sales would be lost as consumers switch to alternatives outside the market. Given the loss in unit sales, the overall profitability of the price increase then depends on the relative size of (a) the resulting profit lost from the loss in sales and (b) the profit gain from charging higher prices on all remaining units. If the proposed market appropriately separates close from distant substitutes, it will pass the test; otherwise, it will fail.
In the HMT, the relevant question is how customers would respond on the margin; i.e., how many customers would change their purchase behavior in response to an incremental price change. In a proposed market, there may be many inframarginal consumers who are less price-sensitive and would not change their purchases in response to a small price increase. However, the market would still be too narrow if a large enough number of other customers would leave the market.
Application to labor markets
On the labor market side, market definition is fundamentally about substitution between employers. As a result, the same HMT framework can be applied for labor market definition. In the labor setting, we consider a hypothetical single employer, or “buyer” of labor services (monopsonist), rather than a single seller of products (monopolist). The HMT then assesses the profitability of a relative decrease in compensation by the hypothetical monopsonist, rather than a price increase.
As in the product case, the profitability of a decrease in compensation hinges on substitution: how many workers would decide to quit and find a job outside the proposed market? Given the loss of workers, the overall profitability of the compensation change depends on the resulting loss of unit sales, the lost profit due to those lost unit sales, and the increase in profit from paying lower compensation to the remaining workforce.
As in the product case, a proposed labor market will only pass the test if it appropriately separates close from distant substitute jobs. And, as it does for product markets, the HMT for labor markets focuses on responses on the margin; even if some workers place a high value on jobs within the market, the market would be too narrow if enough of the workers with close alternatives would decide to leave in response to the compensation change.
It may be instructive to contrast the approach taken to product markets and labor markets. In product markets, we are looking at consumer preferences for alternative products and sellers’ supply of those alternatives. In labor markets, we are looking at worker preferences for alternative employers and employers’ labor demand.
Whether in the product or labor setting, the HMT is applied in different ways depending on the facts of the case and the available evidence. The HMT is a framework; there are different types of evidence or analyses an economist can use to perform the test. However, in any application, the underlying question is the same: what does the available economic evidence say about substitution?
Use of SSNIP tests in the HMT
On the product side, the HMT is often implemented by assuming a specific numerical value for the price change (SSNIP): e.g., a 5% or 10% price increase over the course of a year. Applying a particular value for the SSNIP is meant to add a degree of numerical precision to the HMT.
An economist implementing an HMT may attempt to bring further precision by applying a numerical SSNIP test. In a numerical SSNIP test, the economist applies formulas to calculate the profitability of a SSNIP based on certain economic inputs. For example, in the product context, the test may quantify the profitability of a 5% SSNIP based on price elasticities for firms in the market, diversion ratios between firms, and margins between price and marginal cost. The formulas for a SSNIP test can be derived or arranged in different ways to better fit the set of inputs that may be readily estimated.
A numerical SSNIP test is one way, but not the only way, to evaluate a candidate market within the HMT framework. In some cases, such a test can provide useful precision. For a given formulation of the test, there are various ways to estimate the inputs (e.g., price elasticity or marginal cost markup), depending on the available information. However, the actual precision of the test depends on the availability and precision of the inputs that are used.
To illustrate, consider a specific type of numerical SSNIP test used in the product market context: critical loss analysis using an aggregate diversion ratio. Critical loss analysis involves comparing the actual loss, i.e., the predicted loss in unit sales due to a SSNIP, with a critical loss value, i.e., the loss of unit sales at which the hypothetical monopolist would break even given the size of the SSNIP. If the actual loss exceeds the critical loss, then the SSNIP would not be profitable, and the proposed market would fail the test. With certain assumptions on the nature of customer demand and firm behavior, the test can be formulated as a comparison between the critical loss value and the aggregate diversion ratio (the share of lost units recaptured by other firms in the market, following a price increase by one of the firms in the market). In the product market setting, a proposed market passes this test if the aggregate diversion ratio exceeds the critical loss, which in turn depends on (1) the size of the SSNIP and (2) the markup of price to marginal cost, often approximated using accounting data.
One can derive a similar formula for the labor context. In this case, in place of the price markup, the critical loss depends on the wage markdown, the margin between compensation and the incremental value of the marginal worker to the firm (the “marginal revenue product of labor” in economic terminology).
If we knew each of the variables in this calculation, and if the underlying assumptions about worker and firm behavior were appropriate given the facts of the case, we could use this calculation to add precision to the HMT. However, precise measures of these variables may not be available, in which case the calculation might not add precision to the HMT framework. In particular, for this specific variation of the SSNIP test, we do not typically observe any proxies for the “wage markdown” or “marginal revenue product of labor.” For example, it could be difficult to evaluate how much less customers would buy if a store had fewer cashiers.
Moreover, this specific formula assumes that the monopsonist would only adjust its labor input due to the SSNIP. In reality, it may be profitable to adjust other inputs at the same time. We could generalize the formula to account for changes to these other inputs as the firm reduces its workforce. But this leaves more variables to measure: e.g., how would the store adjust on other margins to compensate for (or accommodate) a reduction in cashiers? These questions may not have precise answers, and in such instances this numerical SSNIP test may not be appropriate.
Regardless of how the HMT is applied, the key is whether the best economic evidence is used to assess worker substitution, not whether a specific numerical SSNIP test is used. The fundamental goal is to identify substitute jobs, and an appropriate application of the HMT uses the best available evidence on worker behavior to do so.
Economic evidence on worker substitution
Economic considerations when applying the HMT
When applying the HMT for labor markets, an economist must consider the full range of options that workers may substitute toward in response to a change in employment terms.
A key insight in labor economics is that the substitutability of different jobs typically depends on the skills required for the job, and on worker preferences for different types of work. Those skills (and preferences) are not necessarily tied to a given industry or occupation. To take an example from FTC v. Kroger Co., a grocery store cashier could typically find work as a cashier in another retail industry, or in a non-cashier role (e.g., retail sales) that requires similar skills.
As a result, defining a relevant labor market is often not as simple as determining the industry in which the firms operate, or employees’ specific occupations in their current roles. Substitution can potentially occur within and between job titles, occupations, and industries, and an economist needs to account for all dimensions of substitution when applying the HMT.
Market definition also involves delineating the geographic scope of competition, as the HMT is applied to a prospective market defined by both geography and product dimensions at the same time. In the labor context, workers may be willing to relocate to accept other positions, or they may focus on alternative jobs near their current residence. As such, geographic competition may be local, national, or somewhere in between.
Common job transition patterns highlight need for careful analysis
Data on job transitions illustrates the breadth of options workers often choose between and the potential complexity of drawing labor market boundaries. Job transition data identifies the alternative occupations and employers that workers choose when they go on the market. Job transition patterns do not necessarily map precisely to substitution patterns, a point that we return to below. However, they are often informative about the range of potential substitute jobs.
Figures 1 and 2 summarize data from the U.S. Census Bureau on job switching across industries. The data tracks individuals as they transition between employers and identifies the industry of each employer according to the standard North American Industry Classification System (NAICS), a commonly used industry typology. Specifically, the publicly available data identifies the ‘sector’ (e.g., ‘Manufacturing,’ ‘Construction’) and ‘three-digit industry’ or ‘sub-sector’ (e.g., ‘Chemical Manufacturing,’ ‘Specialty Trade Contractors’) of each employer.
