Minimum Wages: Short rejoinder to Geloso

A few days ago I posted here at NOL a short comment on some reaction I’ve seen with regards to Seattle’s minimum wage study. Vincent Geloso offers an insightful criticism of my argument. Even if his point is quite specific (or so it seems to me), it offers an opportunity for some clarification.

But first, what was my argument? My comment was aimed at a specific point raised by advocates of increasing minimum wages. Namely, that even if Seattle’s study shows an increase in unemployment, a study with a larger sample may say otherwise. My point is that the way I’ve seen this criticism raised is missing the economic insight of minimum wage analysis, namely that jobs will be lost in less efficient employers and employees first. So far so good. The problem Geloso points out is with my example. I refer to McDonald’s as the efficient employers fast food chain (think of economics of scale) and as less efficient employers the neighborhood family-run little food place (neighborhood’s diner).

Geloso correctly argues that different employers react in different ways. It is expected, for instance, that a larger employer such as a fast-food chain would have more options to make a marginal adjustment when there is an increase in minimum wages. Of course, I agree, but the point I’m rising is about where jobs will be lost first (not the specific mechanism in each employer). Geloso flips my example and argues that a small diner has more (in relative terms) to lose by letting go one out of two employees than a fast food joint that has to let one employee go among maybe ten thousand. By letting one employee go, the small employer loses a larger share of its output. Therefore a small employer would be more inclined to keep all of his labor force and cut costs on another front (less hours work in average doesn’t cut it, that’s like a shared unemployment that would also cut output down).

A large employer like a fast food chain, however, can let one out of ten thousand employees go because the loss in output is not that significant. I have two issues with this example. The first one is that a fast food chain is facing the increase in minimum wage ten thousand times, not two. To cut even the rise in cost, the firm fast food chain has to cut down its labor force 15% (1,500 employees.) But I think the problem with this example does not end here. If it were the case that small diners don’t cut employment but fast food chains do, then we should see more unemployment in larger employers than in small neighborhood diners.

A second point I want to make is with Geloso’s argument that the study is about focusing “like a laser” on one out of multiple channels in the group most likely to respond in that manner (unemployment?). That the study, as long as the focus is on unemployment, should focus on the less efficient employers (and employees) first, and not just look at the unaffected employers because that’s where we just happen to have better statistics for is my point. There are two options. The first option is that what matters is focusing on the channel the increase in cost will be managed by employers. But this is neither a focus on unemployment nor on the criticism I’m replying to. Option number two, that the study should focus on the employers “most likely” to reduce unemployment, which is actually my point regardless of how many “channels” are included in the sample.

Advertisements

Minimum Wages: Where to Look for Evidence (A Reply)

Yesterday, here at Notes on Liberty, Nicolas Cachanosky blogged about the minimum wage. His point was fairly simple: criticisms against certain research designs that use limited sample can be economically irrelevant.

To put you in context, he was blogging about one of the criticisms made of the Seattle minimum wage study produced by researchers at the University of Washington, namely that the sample was limited to “small” employers. This criticism, Nicolas argues, is irrelevant since the researchers were looking for those who were likely to be the most heavily affected by the minimum wage increase since it will be among the least efficient firms that the effects will be heavily concentrated. In other words, what is the point of looking at Costco or Walmart who are more likely to survive than Uncle Joe’s store? As such, this is Nicolas’ point in defense of the study.

I disagree with Nicolas here and this is because I agree with him (I know, it sounds confused but bear with me).

The reason is simple: firms react differently to the same shock. Costs are costs, productivity is productivity, but the constraints are never exactly the same. For example, if I am a small employer and the minimum wage is increased 15%, why would I fire one of my two employees to adjust? If that was my reaction to the minimum wage, I would sacrifice 33% of my output for a 15% increase in wages which compose the majority but not the totality of my costs. Using that margin of adjustment would be insensible for me given the constraint of my firm’s size. I might be more tempted to cut hours, cut benefits, cut quality, substitute between workers, raise prices (depending on the elasticity of the demand for my services). However, if I am a large firm of 10,000 employees, sacking one worker is an easy margin to adjust on since I am not constrained as much as the small firm. In that situation, a large firm might be tempted to adjust on that margin rather than cut quality or raise prices. Basically, firms respond to higher labor costs (not accompanied by greater productivity) in different ways.

By concentrating on small firms, the authors of the Seattle study were concentrating on a group that had, probably, a more homogeneous set of constraints and responses. In their case, they were looking at hours worked. Had they blended in the larger firms, they would have looked for an adjustment on the part of firms less to adjust by compressing hours but rather by compressing the workforce.

This is why the UW study is so interesting in terms of research design: it focused like a laser on one adjustment channel in the group most likely to respond in that manner. If one reads attentively that paper, it is clear that this is the aim of the authors – to better document this element of the minimum wage literature. If one seeks to exhaustively measure what were the costs of the policy, one would need a much wider research design to reflect the wide array of adjustments available to employers (and workers).

In short, Nicolas is right that research designs matter, but he is wrong in that the criticism of the UW study is really an instance of pro-minimum wage hike pundits bringing the hockey puck in their own net!

Minimum Wages: Where to Look for Evidence

A recent study on the effect of minimum wages in the city of Seattle has produced some conflicted reactions. As most economists expected, the significant increase in the minimum wage resulted in job losses and bankruptcies. Others, however, doubt the validity of the results given that the sample may be incomplete.

In this post I want to focus just one empirical problem. An incomplete sample in itself may not be a problem. The issue is whether or not the observations missing from the sample are relevant. This problem has been pointed out before as the Russian Roulette Effect, which consists in asking survivors of the increase in minimum wages if the increase in minimum wages have put them out of business. Of course, the answer is no. In regards to Seattle, a concern might be that fast food chains such as McDonald’s are not properly included in the study.

The first reaction is, so what? Why is that a problem? If the issue is to show that an increase of wages above their equilibrium level is going to produce unemployment, all that has to be shown is that this actually happens, not to show where it does not happen. This concern about the Seattle study is missing a key point of the economic analysis of minimum wages. The prediction is that jobs will be lost first among less efficient workers and less efficient employers, not equally across all workers and employers. More efficient employers may be able to absorb a larger share of the wage increase, to cut compensations, delay the lay-offs, etc. This is seen by the fact that demand is downward sloping and that a minimum wage above its equilibrium level “cuts” demand in two. Some employers are below the minimum wage (the less efficient ones) and others are above the minimum wage (the more efficient ones.) Let’s call the former “Uncle’s diner” and the latter “McDonald’s.” This how it is seen in a demand and supply graph:

minimum wage

Surely, there is some overlapping. But the point that this graph is making is that looking at the effects minimum wage above the red line is looking at the wrong place. A study that is looking for the effect on employment needs to be looking at what happens with below the red line. This sample, of course, has less information available than fast food chains such as McDonald’s; this is a reason why some studies focus on what can be seen even if the effect happens in what cannot be seen (and this is a value added of the Seattle study.)

This is why it is important to ask: “what do minimum wage advocates expect to find by increasing the sample size?” To question that minimum wages increase unemployment, then the critics also needs to focus on the “Uncle’s diner” part of the demand curve. If the objective is to inquire about something else, than that has no bearing on the fact that minimum wage increases do produce unemployment in the minimum wage market and at the less efficient (and harder to gather data) portion of it first.

PS: I have a previous post on minimum wages that can be found here.

Minimum wages: The economist as a psychologist

Both Ludwig von Mises and F.A. Hayek are known for arguing that there is no such thing as a good economist who is only an economist. For these two thinkers, a good economist-as-scientist also needs to know history, philosophy of science, ethics, and physics. Mises and Hayek are thinking of what an economist-as-scientist should be familiar with and have some minimum knowledge beyond his discipline.

I would add that the economist as public educator, that is, when the economist talks as an economist to non-economists, also needs some awareness of psychology. I may not be using the term “psychology” in the most proper way, but I mean the awareness to understand what the interlocutor feels and needs and then figure out how to communicate economic insights in a way that will not be automatically (emotionally or psychologically) rejected; how to make someone accept an economics outcome they do not want to be true. How to break the bad news with empathy? This is a challenge I try to get my students to understand as one day they, too, will be economists out of the classroom in the real world.

A few days ago I found myself unexpectedly in the “psychologist” position. Only seconds after meeting for the first time two persons dropped the question: “So, what do you think about increasing the minimum wages, should we do it?” I knew nothing about these two individuals, and the only thing they knew about me was that I’m an economics professor. The answer to such a question is an Econ 101 problem: if you increase the minimum wage (above the equilibrium price) some lucky workers will get a wage increase at the expense of other ones loosing their jobs.

The first question I asked myself was “what do these two nice ladies actually want, the analytical/scientific answer, or do they want instead the ‘professor’ to confirm their bias?” This might be a delicate discussion since they may well have a loved one in the minimum wage market.

The first thing to get out of the way is that my answer as an economist is not ideologically driven or does not respond to secret political agendas. How can that be made clear? One way is to show the economics profession’s consensus on the subject from an impersonal position. I explained to them that any economics textbook from any author from any country in the world used in any university would say the same thing: “If you put the price of labor above its equilibrium (a minimum wage), it will produce a disequilibrium (unemployment). You cannot fix the price outside equilibrium and at the same time remain in equilibrium.” Yes, as Ben Powell reminds us, even Paul Krugman agrees on this. By mentioning a worldwide consensus there is no room for ideological or political agendas. It is important to mention that economics is not always about politics. The economic analysis of minimum wages has nothing to do with being a Democrat or a Republican; the political position of each party may differ, but those are not economic analyses, those are political strategies.

The next step was to deal with the issue that if such a consensus exists, why are there mentions of studies showing no harmful effects of increases in minimum wages. This is no mystery either. A well known reason of why an increase in minimum wages does not increase unemployment is because in fact there is no such increase. The politician may say he is increasing the minimum wage, but he does not say that the minimum wage is being located just above the equilibrium level and therefore he is not doing much. Another reason is to look at the effect of a minimum wage increase in a small location where low skilled workers can get another job in the next town without the need to move and therefore they will not show up as unemployed. This is another case of an ineffective increase in minimum wages. Or maybe the minimum wage increases but a benefit goes down. The total compensation to the employee does not change, its composition does.

But can the economist show his claim? Is there more clear evidence that the effects of increasing minimum wages does do harm than the complicated cases where there are no harmful effects? Again, I went geographically large. First, I compared the U.S. with Europe, which has higher minimum wages with respect to the U.S. In Europe you find higher unemployment rates, higher unemployment in the youth population, and also higher long-term unemployment. Second, I brought the case of the U.S. Fair Labor Standard Act of 1938, which fixed the minimum wage at 25 cents per hour. This law included Puerto Rico where the many workers were earning between 3 to 4 cents. Bankruptcies and unemployment skyrocketed. It was in fact unions who asked Congress to make an exception for Puerto Rico, which took two years to consider. For two years people in Puerto Rico were forced to work in the black market or fail to make a minimum income. Want more cases? See here. “See,” the psychologist says, “minimum wages are very dangerous; you can seriously harm yourself. Is that a bet you really want to make?”

Two issues remain to be explained after dealing with these three problems, (1) objection to minimum wage increases is not (necessary) an ideological or political position, (2) studies that deny the effect are doubtful for easy reasons to understand, and (3) if you look at a sample broad enough the economist’s prediction is right there.

First, when an economist objects to an increase in minimum wages it does not mean we do not want wages to go up. We are just saying that a minimum wage increase is not the right way to do it. I wish it were so easy, but the laws of demand and supply inform otherwise. I guess most economists would advocate for minimum wages if their negative effects were not real. Second, explain Milton Friedman’s lesson that a policy is valued by its results, not by its intentions. The economist objects because of the unintended effects of fixing the price of labor outside equilibrium, not because we wouldn’t like to see real wages increasing.

On Minimum Wages, Health Code Violations and Proper Assessment

A few days ago, Tyler Curtis at the Freeman (the Foundation for Economic Education’s flagship publication) posted a short piece on the minimum wage and health code violations in restaurants. Curtis based on a paper (unavailable in full) by Srikant Devaraj who asserted that increases in minimum wages in Seattle had led to increases in health code violations by restaurants (heavily affected by the increases).

Devaraj used a standard difference-in-difference econometric approach. The problem underlined by some was the choice of benchmark for the method : New York City. New York City and Seattle are two very different cities with different health codes. It is hard to make this claim stick even if the uncontrolled results (before statistical tests) show an increase in health code violations. Nonetheless, I have been able to find one other study that shows – based on Californian data – that there was a very small deterioration in health code violations (especially by the top restaurants) following increases.

Now, I am not convinced by the econometric design of both, but I am axiomatically convinced. This is where I think economists have made the error of relying too much on empirical methods. While, as an economic historian, I always favor more data, I also am trained to be skeptical about data does not say.

In the case of the minimum wage, the debate has raged between economists over the employment effects (i.e. the demand for labor). But this is a fraction of everything involved with the production of goods in industries affected by minimum wages. For an employer, a cost is a cost regardless of the form it takes. If an employer is forced to pay somebody above what they produce in value, then something has to give. For a 5% increase in the minimum wage, it is doubtful that an employer with three employees will be willing to sack one third of his workforce (and roughly one third of his output). So, he can cut costs differently. He may ask employees to buy their uniforms, he may refuse to provide them with free lunches, he may also even decide to cut on quality of his service – as is the case with the two studies outlined above.

The problem is that no study of the minimum wage has attempted to measure all these effects at once!  There is no study that looks simultaneously at hours worked, people employed, type of people employed (substitution effects), prices for consumers, quality and marginal benefits (uniforms, free lunch, insurance, etc.) on both the short and long-terms levels and trends (they also rarely adjust the minimum wages for regional purchasing power parities and the under-reporting of tips)

The health code violations papers show how many channels employers can use to adapt – channels which some fail to account for when they proclaim that we can raise the minimum wage without adverse consequences. Maybe its time that we, as economists, try to be more cautious when we make claims about the minimum wage’s minimal effects.

Minimum wage and length of poverty spells

I had a pre-programmed blog post on the issue of the minimum wage and poverty which was preempted by Mark Koyama (a blogger here at Notes on Liberty). The tweet is below and it has forced me to adjust the post.

Mark is absolutely right! Let me explain why with my own spin on it.

First of all, the demand curve slopes downwards – always. However, the method of adjusting to price changes (wages are a price and the minimum wage is a price control) is not an empirical constant.  I am unlikely to fire workers for a 1% in the inflation-adjusted minimum wage. Firing workers implies transaction costs that are dependent of context (for example, if I am friend with my employee, this is a transaction cost in the form of a lost friendship), firm size (I won’t fire my only employee which represents 50% of my output for a 1% hike in MW) laws (firing and hiring regulations), institutions (social institutions, reputation, norms), my clientele (how elastic is their demand) and technological alternatives. For a 1% increase, I am likely to reduce work hours or cut marginal benefits (no free soup for you). For a 10% increase, I am more likely to consider the option of firing a worker or I may shift to a new technological set that reduces my demand for labor.  It may happen rapidly or take some time, but there will eventually be an adjustment.

In any case, the minimum wage will imply some losses with a deadweight loss. Only the method by which it materializes is debatable.  By definition, some people will be hurt and generally and even if supply is super-elastic (doubtful), some suppliers (workers)  will be ejected from the market (or the quantity of labor they can supply will be ulitmately reduced). Since the minimum wage generally tends to fall on unskilled workers, this must be correlated with workers close to the poverty line.

Ideally, we’d need a measure of the minimum wage to be compared with the “at-risk” population over a long period of time in order to encapsulate all the effects of the minimum wage. The perfect measure is the “length of poverty spell” variable which has been emerging progressively from the BLS. The problem is that it is not broken down by state. Fortunately, Canada has that variable (well, a low-income variable which is a relatie poverty measure) for provinces. Inside the Survey of Labor and Income Dynamics (affectionally known as the SLID), this longitudinal variable has a span of eight years. Basically, we can know if a person has been below the low-income threshold for up to eight years. Let’s take that extreme measure and plot it against the minimum wage divided by the average wage.

As one can see from the scatter plot below, there is a more or less clear relationship between the minimum wage as a share of the average wage and the length of poverty spells. What is more impressive is that this graph is not a regression. More precisely, the provinces with the highest minimum wages (like my own province of Quebec and the province of Nova Scotia) also have the most extensive social welfare nets. Alberta,  a province with the lowest minimum wage ratio and one of the least “generous” social welfare net in Canada, is at the very bottom of the pack in terms of the persistence of poverty.

minimumwagepoverty

I think this graph acts in very modest (but clear) support to Mark’s point (which is also the point of Burkhauser, Sabia, MaCurdy and many others)

On quasi-rents and the minimum wage

Ryan Murphy of Southern Methodist University has a new article published in Economics Bulletin regarding the minimum wage and “quasi-rents”.  The argument made by Ryan has the advantage of theoretically fleshing out a point made by many skeptics of the new literature. Generally, the argument has been that in the short-term, the minimum wage may have minimal effects, but in the long-term, firms will adjust.

I tended, until Ryan’s article, to be more or less skeptic of the value of this counter-argument. My point has always been that the new literature (like the Dube-Lester-Reich paper) tends to act as a partial equilibrium story (focusing only on one sector only or one indicator). My view has always been very “Coasian” in the sense that there are transaction costs to adapting to any new minimum wage rate.

The height of the hike and what industries are primarily affected will determine the method of adjustments. Firms can cut on benefits, substitute between forms of labor (the minimum wage increases the supply of older workers which remplace younger inexperienced workers), hours or training. They can also, depending on the elasticity of demand for their products, increase prices or cut quality. They can also cut employment. All of these are channels of adjustment and they will be used differently depending on the context. They are all different expressions of the fact that the demand curve slopes downward. But each expression has costs to be used that are to be weighted against their benefits – which are highly circumstantial. For example, if I have a firm of two employees, I will not sacrifice half my workforce by firing a worker (thus sacrificing 50% of my output) for a 5% hike in the minimum wage. Not only would this be an over-reaction, but there are transaction costs for me to fire that worker : separation fees, emotional pain, learning what the employee was doing etc. Reducing his hours would be a safer adjustment.

Until there is a study that measures all of these adjustments channels at once, I am skeptical.

So where does Ryan’s story come in? Well, none of my arguments had a long-term component. They were largely void of any time dimension. While I am aware of research like those of Meer and WestClemens and Wither and Clemens regarding job growth patterns following minimum wage hikes, I always discounted that argument. I was always reluctant to engage in long-term reasoning because I felt it was conceding a point that ought not to be conceded even if that counter-point is valid.  I only used it to top up the rest of my argument. But Ryan introduced to me the concept of quasi-rents, of which I had vaguely heard during my undergraduate microeconomics class.

Basically, here is the argument about quasi-rents: in the short-term, there are rents to be extracted from fixed factors of productions. Firms need these quasi-rents to remain in business, but only in the long-run.  However, if labor can find a way to capture the rents in the short-run, they will get higher earnings and employers will not fire people as much. As a result, there is basically a reshuffling of the consumer surplus. However, in the long-run, nothing is fixed and firm owners can adjust by shifting to different production methods. Thus, they will reduce their future hirings. In Ryan’s words:

But the on-impact negative effects of minimum wages may be hidden. In the longer run, after the quasi-rent is dissipated, the owner would have the incentive to eventually switch from more labor-intensive methods to ones that are less globally efficient (this being the conventional “demand slopes down” result). More perniciously, the threat of future increases in the minimum wage may create regime uncertainty undermining a willingness to invest in the types of technology and capital complementary to low skilled labor, thereby reducing employment for low skilled workers. That is to say, the risk of the appropriation of quasi-rents can shift investment towards capital unlikely to be appropriated via the minimum wage. Repeated and arbitrary increases in the minimum wage worsen this risk. This is consistent with the recent shift towards long run effects of increases in the minimum wage, for instance Meer and West (2016).

This is exactly what Andrew Seltzer found for the introduction of the minimum wage during the Great Depression in certain American industries. In the short-term, the capital was more or less fixed and production methods could not be abandonned easily. In the long run, firms adapted and shifted production methods. This is why Ryan’s argument is convincing. It offers a theoretical explanation for the empirical results observed by Dube, Lester and Reich or Card and Krueger. It fits well with theories of imperfect markets (damn I hate that word that is basically saying that all markets have frictions) like those of Alan Manning (see his Monopsony in Motion here).

This is the kind of work on the minimum wage that, if measured, should force considerable requestionning on the part of minimum wage hike advocates.