Wednesday, August 5, 2015

The Minimum Wage, Sunk Costs, Returns, and Capital

I have previously thought through the implications of minimum wage hikes on an industry through a finance lens.  In the end, going through that process has brought me to the conclusion that, (1) markets with relatively open access at an aggregate level and uncertain but persistent returns on investment can be competitive in the aggregate while individual firms appear to have excess profits (reflecting monopoly or monopsony rents), (2) the effects of MW increases on the labor force will eventually play out through capital destruction and reallocation, so that the employment effects may not have a shape that is particularly different from other shocks to labor utilization, and (3) there is a "part of the elephant" element here, where you would find no effect, disemployment, or even employment growth, depending on what part of the market you measure, even though we would expect net employment loss, or more precisely some employment dislocation and capital destruction.  Some of that loss - maybe even most of it - would come from outside the industries directly involved in minimum wage labor because the net effect would result from capital reallocation, and some of that capital may be reallocated out of other industries, into industries that have eliminated minimum wage workers.

So, the only way to fully capture the employment effects of a MW hike may be to look at the total labor market, but the effect at that level will frequently be too diffuse to measure with significance.

If this is confusing, please check the link above.  That post is probably a prerequisite for this post.

A narrative version of an industry we might think about would be gas stations.  We might imagine two stations at an intersection.  When they originally locate, both have equivalent expectations of their returns on capital.  But, once they locate there, traffic patterns will emerge which were not originally predictable.  These might favor one station over the other.  So, one station owner will experience persistently higher returns than the other one.  Another station could locate on another corner at the intersection, or at a nearby intersection.  In this way, the market for gas stations would be competitive, even though these stations have persistently different outcomes.  So, the station with the better returns would appear to have monopoly profits, but really they are balanced out by the low returns of the other station, so that the average returns for the two stations would still be a reflection of competitive expected returns for this type of investment.  And, the occasional failure of the weakest firms creates survivorship bias that makes the profit level of the entire industry appear to be higher than competitive returns would be, even though the profit on all of the invested capital would not.

This is basically portfolio theory applied to an individual industry.  At the level of equity portfolio management, we all understand that some stocks perform better and some perform worse, but that the average returns to equities over time reflect some underlying required return that the marginal investor demands.

Here was the final graph from that earlier post.

The end result was a loss of jobs at the weakest firms and an expansion of the remaining firms plus an inflow of capital from other industries to re-establish a new equilibrium that re-attains the original expected ROI of the industry for marginal new investments, at a slightly lower total aggregate employment level.  This means that a small number of firms in the industry fail, a large number of firms in the industry expand slightly, and some capital is reallocated from other industries into this industry to replace some of the lost firms.

Using our gas station narrative, several weak stations may fail, the remaining strong stations may add pumps or workers when they see added traffic as a result, and a few new stations that use self-serve kiosks may open up.  These new stations may have largely been uncompetitive when the market included weak firms with low wage employees, but now with fewer stations and higher wages at those stations, these stations would be more competitive.  Some observers make the claim that this inflow of capital is a positive outcome of minimum wage hikes, but I think this is an error.  The MW hike can't create capital, it only causes capital to be reallocated from some other use.  So, we now have a gas station with kiosks instead of a station with an attendant.  But, the creation of that station means that we don't have a car wash that the new station's owner had intended to open before the MW hike changed the competitive context.  Essentially, we have destroyed capital that was being profitably deployed in the beginning context (the weak station).  So, the MW heavy industry may be slightly more productive now because of the new capital, but some other industry is less productive as a result.  The net post-MW hike productivity must overcome the destruction of the capital in the lost station, so it seems to me that the net effect of the MW hike is very likely to hurt aggregate productivity.  We used the same amount of capital, but now we don't have a car wash and we do have some unemployed former gas station attendants.

Regarding the question of whether capital is a substitute or a complement to labor, it seems to me that the sorts of capital adjustments created by MW hikes will make capital more of a substitute for labor, lowering labor's total share of production.  And, in my narrative example here, we do see that the higher MW doesn't affect the total amount of invested capital or the return to capital (Losses to the weak station are countered by gains to the other station and the new station.) but it does lower total production (There is no car wash.) and employment (the employees of the failed station).  The loss in production comes from the decline in labor.

Thinking through all of these changes we would see:

1) A loss of employment across the wage scale at the few firms which fail.

2) Higher wages specifically for the minimum wage workers at the remaining firms, representing almost the entire industry.

3) Higher prices at the remaining firms, which would appear to both observers and industry managers as a result of "passing the costs on to the customer" as the result of some sort of market power, but in the aggregate would really just reflect the new context introduced by the changing cost structure and marginal loss of competitors.  Even in a perfectly competitive industry, I think there would be some change in the mix of price and quantity, simply reflecting various elasticities.  In effect, this is just another way of saying that the destruction of capital leads to lower real incomes across the economy (and one way that change is conveyed, assuming stable nominal incomes, is through higher prices in the affected industry.)

4) Some marginal level of new capital would be attracted to the affected industry, leading to some marginal addition of new jobs in the industry (likely at higher wages because the failure of some firms with low wage business models may allow the introduction of business models that didn't depend on low wage structures) and some loss of new employment outside the industry because of this capital reallocation, which would be very difficult to notice and would include an indeterminate range of wage levels.

Regarding points 3 and 4, it seems as though, if demand for the given product is very inelastic, prices will rise substantially and capital will be reallocated into the affected industry, so that the loss in total consumption would come from outside the industry.  If demand is very elastic, production and employment within the industry will fall, and capital will be allocated outside the industry, as it would have without the MW hike, and the loss in total consumption would come from a decline in the production of the affected industry.

So, there would be a dislocation of labor and a loss of productive capital.  But, we could see an inflow of capital into the industry, profits at remaining firms apparently bolstered by higher prices, the vast majority of MW workers at the time of the hike keeping their jobs at higher wages, and possibly even expansion at some remaining firms.

So, I think there are a couple of implications here.  One is that, even though the net effect would be negative in the short run (through the dislocation) and the long run (through capital destruction), there would be many ways to measure significant positive outcomes.  But another is that this policy may not have implications that are especially bad for the most vulnerable workers.  The job dislocations may be spread throughout the wage spectrum, so that most of the gains go to the vast majority of MW workers who keep their jobs, and most of the dislocation falls on non-MW workers who lose their jobs and on consumers through higher prices within the affected industry and through the loss of new production outside the industry.

Maybe, compared to the way I have thought about MW hikes, the negative effects fall on the most vulnerable workers in the same way that negative effects always fall most harshly on the most vulnerable, not because the negative effects of the MW themselves end up being focused on vulnerable workers.

In effect, all of the seemingly contradictory findings may be true.  Maybe, on net, MW hikes damage the labor market as a whole to the benefit of most existing MW workers.  So, maybe the debate about the policy is not particularly unique.  Maybe this creates some persistent downward redistribution, some dislocation, and some loss of growth, and the question is about the whether the scale of those effects is positive, on net.  I tend to come down against dislocations and lower growth, but I might think twice about assuming that the negative effects of MW hikes are absorbed by low wage workers particularly more than they are in other interventions that generally reduce growth or create labor dislocations.

Of course, one other factor that I haven't considered here is the extent to which demand for entry level workers is permanently depressed or that some very low productivity workers are permanently locked out of employment.  There are unfortunate cases where special labor arrangements for physically disabled or learning disabled workers are made illegal by these laws, or where labor markets in places like Puerto Rico or American Samoa are especially hurt.  As a national minimum wage level increases, the range of places that would be especially hurt in this way would expand.  It is probably important to keep in mind how strongly income levels correlate with labor force participation.  Here is a great interactive map from the Census Bureau that really drives that point home.  We fret over aggregate changes in labor force participation of a few percentage points, but the difference in LFP between neighborhoods can be very sharp.  (Click on the link for a larger version.)

In the long run, workers are still going to tend to acquire work with wages proportionate to their productivity.  Even if that is an extremely inefficient process, the consequences of lower broad levels of labor productivity will tend to accrue to the most vulnerable and least skilled.  The fact that these jobs tend to have higher turnover than higher wage jobs accelerates that process.  As with most policies, it seems that "We are the 100%." holds.


  1. That's a good point, too often it's assumed that the capital investment created by higher labor costs must raise living standards generally, but there's really no reason that should be true.

    The LFP map is interesting.

    The only solid arguments I've seen for MW hikes are dependent on the effects of other government actions, particularly the high MTRs created by welfare and the high costs associated with regulation -- i.e. you probably don't want to live nearby people making less than $1/hour, and indeed it may not even be possible for them to comply with regulation (e.g. housing/automotive/food-quality codes) at those incomes, and it may not make sense for them to work at those incomes given the MTRs and leisure opportunities created by welfare.

    1. Yes. It definitely is important to think about de facto MTR's, etc. It seems like some combination of work subsidies and guaranteed incomes work best.

      As far as $1/hour wages go though, it looks to me like the natural minimum wage in the US is somewhere around $6 or so. When the minimum wage was at $5, the BLS estimated very few workers at that level.