# The Politiconomist

Where Politics and Economics Hang Out

## Can Circular Effects Justify a Wage Hike?

I couldn’t help but notice Ben Studebaker was doing one of his old saws about how circular effects can justify a minimum wage hike.

I know, I shouldn’t get sucked back into this. But I am. I have a new perspective I want to come at it with, and I was actually planning a similar post before Mr. Studebaker posted his. Down the rabbit hole we go!

I’m going to make a simple point in a very exacting way. The heart of my claim summarizes neatly: If the problem is a broken link between consumption and wages, you cannot fix the problem by trying to exploit that link.

I will first look at the underlying accounting model Mr. Studebaker employs. I’ve never been one to leave out explicit accounting equations when doing accounts, so I will develop it with a Social Accounting Matrix (SAM). This is nothing more than addition. I will show that it is impossible for the problem to be strictly a private sector model and still comport with the facts at hand.

I will then show that Mr. Studebaker’s analysis of the 1950s and 60s economies holds even when we include government accounts.

Next I will move onto an open economy model to establish the plausibility of other explanations. This, of course, only shows Mr. Studebaker is wrong to assume a tight consumption/compensation link and suggest alternatives. I will have to link this to his empirical considerations to evaluate which model we ought take.

I will then wrap together his empirical points as well as a few of my own to show that the problem rests very much in the Current Account. The money available for wages is very likely being sent abroad. I acknowledge this isn’t a perfect model, and I offer some other accounts to look for the difference in, but I think the case for it is compelling.

Finally, I’ll put the New Labor findings in a wider context so as to more fairly judge them. It is my position that both the statistical methods and conceptual explanations for those findings are flawed, as outlined above.

### A Compensation-Consumption Model

Let me turn your attention to an early claim of Mr. Studebaker’s:

As productivity rises, the economy is able to supply more and more goods and services with the same number of workers. However, increases in the capacity of the economy to supply goods and services are not sufficient to grow the economy–to do this, consumers must be able to purchase those additional goods and services. Until the 1970′s, consumers were able to do this because their wages rose alongside productivity. Instead of firing workers or reducing their hours and salaries, businesses in the 1950′s and 1960′s increased wages and worker benefits, allowing workers to purchase the goods and services they were producing. As a result, the economy expanded quite rapidly during that period. Productivity and wages helped one another in a virtuous cycle.

(Emphasis mine.)

Broadly I agree, but I’ll take exception to the bolded part when the time is ripe.

As a simple thought experiment, let’s imagine we resided in a closed libertarian economy. The macroeconomy is divided into three sectors—households, firms, and investment. Each pays the other. This is traditionally shown as a so-called Societal Accounting Matrix. As I’m presenting it, you read it as [row] pays [column] [entry].

$\begin{array}{c|ccc} & H & F & I \\ \hline H & & C & S \\ F & W & & \\ I & & I & \end{array}$

For example, Households pay Firms Consumption, C. W stands for Total Compensation, I for investment, and S is Household Savings.

The beauty of the SAM is that rows must equal their corresponding columns. There is no great mystery to this; by accounting identities everything bought must be paid for. This is the essence of the argument Mr. Studebaker is making; supply must meet demand*. Firms obviously cannot claim any payment on things not bought; consumers cannot buy things with wages unpaid!

We find ourselves facing three identities:

$\begin{array}{l} W=C+I. \\ W=C+S. \\ I=S. \end{array}$

Notice that from the first equation we can deduce that production (Y=C+I) is equal to compensation. Households drive the economy. Mr. Studebaker is unequivocally correct in a closed libertarian economy about consumers driving the economy.

The trouble is that productivity is locked into compensation! The break in compensation from productivity that he presents, as a matter of accounting, is impossible without other sectors of the economy! I will make the case for this in a moment.

But first, let’s tackle another of Mr. Studebaker’s claims:

The economy attempted to make up for the absent wage growth by making it much easier for consumers to borrow money. Instead of paying for goods and services with wages, consumers paid for these things with credit, which took the form of household debt:

Now, if we look back to our accounting identities, this is a conceivable argument. Borrowing for consumption must be credited to Consumption. (I mean, accounting tautology is a tautology.) In order to keep wages from rising proportionally, we must then debit savings. The trouble with this is that savings (investment) has increased empirically:

Granted, this data is not from a closed libertarian economy. Granted, Personal Savings—the account that better corresponds to the question Mr. Studebaker asked in a more open economy—has fallen with Compensation.

But that’s the point. We can also take this as a partial analysis model of an open economy with a government. We can assume these are the only accounts able to move, as Mr. Studebaker’s analysis implicitly does.

The logic Mr. Studebaker employs does not bear out either the claim that wages fall with falling productivity or that their fall necessitates more borrowing unless we include more accounts.

*The argument that supply creates demand is a stronger one. Instead of supply equaling demand, it says any increase in supply will be met with a increase in demand; we now know there are reasons to believe this is generally not true. Despite often claiming I fall back on Say’s Law, this is precisely what he is doing!

### Adding in a Government

More complicated SAMs are, well, more complicated. As the number of sectors increases, the number of sub-accounts tends to increase faster, meaning that we have to make behavioral assumptions to make claims. This comes with the downside that behavioral assumptions lack the tautological umph! that comes with just being able to point to a simple accounting truth.

Still, we can still see some similar consequences from a 4×4 SAM with a government:

$\begin{array}{c|cccc} & H & F & I & G \\ \hline H & & C & S_H & T_H \\ F & W & & & T_F \\ I & & I & & \\ G & G_H & G_F & S_G & \end{array}$

There are now 4 accounting identities:

$\begin{array}{l} W+G_H=C+S_H+T_H. \\ W+T_F=C+I+G_F. \\ I=S. \\G+S_G=T. \end{array}$

Now, these equations more or less describe the state of affairs pre-1971 when the economy was much closer to closed*. Regardless, they offer a theoretical insight into how the state effects accounts without being muddied by international effects.

Mr. Studebaker’s proposal is that an increase in compensation suggests an increase in consumption is still plausible from the first two equations. I would like to point out the identities it does not necessitate it! An increase in wages could be balanced by any of the other accounts, including state ones.

However, the claim is borne out in the data he puts forward. I’m inclined to agree: prior to 1971, compensation increases meant consumption increases. The first paragraph I pulled above I believe is a fair analysis of that state of affairs.

*I will explore this in more detail in the next section. The barriers to trade in that era came down in 1971, offering an excellent natural experiment.

### International Accounts

Yep, it’s time for a 5×5 SAM. We’ll add in an A account for Abroad.

$\begin{array}{c|ccccc} & H & F & I & G & A \\ \hline H & & C & S_H & T_H & \\ F & W & & & T_F & \\ I & & I & & & NL_P \\ G & G_H & G_F & S_G & & NL_G \\ A & NX_H & NX_F & & & \end{array}$

This gives us 5 accounting equations:

$\begin{array}{l} W+G_H+NX_H=C+S_H+T_H. \\ W+T_F=C+I+G_F+NX_F. \\ S=I+NL_P. \\ G+S_G+NL_G=T. \\ NX=NL. \end{array}$

This is where things start to get very interesting.

It remains a possibility that a consumption increase will end an increase in compensation. But Mr. Studebaker’s own evidence suggests this has not been the pattern of things since 1971.

In my first quote from his piece, I bolded the part about consumers. As it turns out, there is a whole host of possibilities. If compensation increases any of the following things must happen: an increase in consumption, household savings, household taxes, total investment, government subsidies to firms, or net exports by firms; or a decrease in government subsidies to households or on taxes on firms. (All of this is delightfully confounded by the fact that there is a lot of noise from other trends and policy changes.)

Let’s look at where we’re at. Neither a simple Compensation/Consumption model nor a more nuanced one involving the state is sufficient to explain the post-1971 trends. Further, there are a deluge of possible suspects present in an open economy model.

### The Narrative in the Facts

Mr. Studebaker is very fond of the following graph:

For what it’s worth, I’m also very fond of this graph. I’ve written about it before—both independently of this ongoing discussion and in a previous response. It shows, quite clearly, that something happened in 1971 to cause the already drifting wage-productivity relationship to unhook. There’s no mystery to all of this!

That’s the year that Nixon ended Bretton Woods in favor of a more open regime. Prior to that, currency revaluations kept the Current Accounts of nations much closer to balanced. A look at the data shows that the Current Account wasn’t completely balanced pre-1971, but it was much closer to 0:

I want to draw your attention to a few similarities between this and the wage trend, but with a word of caution. Eyeballing time series data is doubly dangerous. Eyeballing any data opens you up for bias and false pattern spotting. Further, seeing lagged effects is more or less impossible. Still, both follow a pattern that I think merits some examination. Post-1971, both briefly rise and then fall together. (I’m not sure why the Current Account briefly went positive after the fall of Bretton Woods; it seems we were a significant lender to nations hit by the fallout?)

There’s also a period in the 90s where this hypothesis fails. I’ll print the graph momentarily, but this is not fully explanatory. Wages aren’t simply going abroad—which I’ll admit I’ve previously represented. While the R2 if calculated isn’t likely to be terribly impressive, the graph between wage share (vertical axis) and balance of trade as a fraction of GDP (horizontal axis) makes me think it’s part of a good model:

I’d be comfortable doing estimations on the fly with a rule of thumb along the lines of:

$\frac{W}{Y}=\frac{CA}{Y}+0.48$

but I’d also clearly be missing something. Those clusters indicate some other variable is shifting around.

Again, without doing any formal statistical analysis, we can look at the Personal Savings account against the wage share. This is pretty weak, though I wouldn’t be surprised if it fit into a more complicated model robustly:

We find a much closer relationship—though, it’s hard to say how close without firing up R—between government subsidies to persons and wage share:

The point being, a lot of things are at work here, and they may well be working with and against each other.

How would I summarize these findings?

1. The fall in the wage share cannot be a direct issue of the private, domestic market.
2. Current account imbalances, including those driven by the deficit, are contributing to the wage shift. Because of the sharp change in the accounts following the demise of Bretton Woods, they are the single clearest explanation for wage decline, but they cannot be the whole story.
3. Government subsidies and personal savings also have something to do with it, but teasing out causation is a bit harder. I’m inclined towards the argument that says that, at very least, government subsidies are helping keep wages low by keeping wage-earners from having to push for higher wages on the supply end and sending wages abroad on the demand end. I’m inclined to view the much weaker personal savings relationship as an obvious corollary of stagnated wages and rising debt, but I’m open to other ideas.

### Wait, So What Happens if We Raise the Minimum Wage?

A full detailing of my thoughts on the state of the relevant empirical literature has been published, but it summarizes nicely enough. I don’t buy that it’s appropriate to control for being a state which raised the minimum wage and then report no correlation when the minimum wage is raised. Of course not—you implicitly controlled for it. This means I fall on the side that says there is a significant decrease in employment (in terms of wage-hours) following a minimum wage hike.

To be fair to Mr. Studebaker, those controls have a respectable acceptance among a minority of economists, but let’s not overstate his position. Most studies, including some which employ the New Labor methods, find that employment falls as the minimum wage rises. For a thorough, 150 page study detailing of the state of the literature, have this bit of light reading. The money quote starts at the bottom of page 114:

This wide range of estimates makes it difficult for us to draw broad generalizations about the implications of the new minimum wage research. Clearly, no consensus now exists about the overall effects on low-wage employment of an increase in the minimum wage. However, the oft-stated assertion that this recent research fails to support the traditional view that the minimum wage reduces the employment of low-wage workers is clearly incorrect. The studies surveyed in this paper lead to 91 entries in our summary tables (in some cases covering more than one paper). Of these, by our reckoning nearly two-thirds give a relatively consistent (although by no means always statistically significant) indication of negative employment effects of minimum wages—where we sometimes focus on results for the least-skilled—and fewer than 10 give a relatively consistent indication of positive employment effects. In addition, we have highlighted in the tables 20 studies that we view as providing more credible evidence, and 16 (80 percent) of these point to negative employment effects. Correspondingly, we have indicated in our narrative review that, in our view, many of the studies that find zero or positive effects suffer from various shortcomings.

To whit: the lit’s there for Mr. Studebaker, but it’s not a robust majority in any sense. And it goes on for several pages detailing specific trends that are very damning to any claim of New Labor ascendency. So, while for much of the preceding I’ve just assumed a compensation hike, it seems that may have been far too generous to Mr. Studebaker’s position.

And that follows from more careful accounting and the evidence he provided.

If the accounting doesn’t speak to you, let me pose it in a more essential way. Mr. Studebaker’s explicit position is that the wage-rate is not necessarily linked to either production or consumption, and presents evidence to this effect. But, he asserts that an increase in the wage-rate would, through this same broken link, increase both compensation and production as a whole.

It simply doesn’t follow.

Written by R. A. Stark

April 10, 2014 at 12:00 PM

### 4 Responses

1. A few thoughts:

The fall of Bretton Woods certainly plays a huge role–by opening up wages to international competition through the huge expansion of international trade that accompanied it, it pushed wages downward. This increase in international trade corresponding with a fall in growth rates in the affluent states. Many economists have argued that imports have also driven down the prices of goods and thereby made it easier for consumers to purchase more with the same wages, but the increase in household debt during the same period suggests this has not been sufficient. To truly restore the wage-productivity relationship, states would need to reconstruct much of Bretton Woods, including the system of capital controls.

Nevertheless, a majority of every dollar spent in the United States continues to contribute to the US economy–most domestic consumption does not go abroad. We are much more globalized than in 1971, but globalization is far from complete. We’re somewhere in between 1971 and wherever it is we’re presumably going in the absence of Bretton Woods. Now you’re right that many studies have not found my connection between wages and consumption, but these studies often take place outside the confines of the liquidity trap, in which purchasing is not the limiting factor on growth because private sector debt is successfully compensating (for the time being) for stagnant wages. Every time anyone has run a minimum wage study between 1971 and 2008, this has skewed the outcome. So long as the economy can lend people money rather than pay them wages, an increase in wages only leads to a decrease in borrowing and there is no net consumption effect.

In the context of the liquidity trap, of the post-2008 economy, there is a lot of money in the United States–not abroad–that is not presently doing anything. You’ll recall in my piece that I pointed to the huge profits firms have experienced post-2008. Those profits are not translating into new jobs or other kinds of investment in production because there is no demand for the goods and services that new production would provide. As a result, companies are sitting on large piles of money that are not playing an active role in the economy. They are throwing much of this money into the bond market, depressing interest rates on 10-year bonds.

Now, I presume you would agree with me that if the state decided to embark on a major stimulus program in which it took this idle money from companies in exchange for bonds and then redistributed that money to poor Americans, those poor Americans would spend the money, consumption would increase, and growth rates would rise, even though this policy would take funds out of the hands of firms. The IMF’s research indicates a multiplier of 1.5 in this scenario, and fiscal stimulus is not particularly controversial in macroeconomics.

What I’m saying is that there is functionally no difference under current economic conditions between taking the idle funds through state borrowing and taking the idle funds via a minimum wage hike. Since fiscal stimulus is political unfeasible, a minimum wage hike is a highly desirable alternative way of moving the idle funds back into the economy.

This argument would break down if we were in a different scenario in which firms weren’t sitting on lots of money and in which consumption were not the limiting factor for growth. Because both of these conditions holds, my argument holds.

Benjamin Studebaker

April 10, 2014 at 12:42 PM

• Your first paragraph has far more of a bite than you seem to realize, and I’ll try to demonstrate that.

Your presumption in paragraph 4 is essentially correct; I even take that IMF study as my preferred one for the most recent recent recession. Empirically, it is reasonable to assume that credits to Government National Accounts will result in 1.5 times the growth of GDP, and vice-versa for debits. This is a logical theoretical consequence of the Income Hypothesis you allude to in paragraph 2. This is the case for fiscal stimulus.

All of which is a complete non-sequitor, despite your claim at the opening of paragraph 5. Compensation increases (never mind if you can expect such a credit from a hike in the wage-rate!) are emphatically not credits to the Government accounts! We cannot apply the IMF’s estimated multiplier if it is not between government spending and output because that is all the IMF’s multiplier covered.

I think you misunderstand why the liquidity trap scenario is convincing in the government spending case in the first place. That money is being used by firms to invest. When the economy is not liquidity trapped, government spending makes the money available more scarce and raises the price of borrowing it. This crimps borrowing and therefore investment. There is a cancellation (not necessarily sum-zero) between government spending and investment.

All a liquidity trap is is the case where virtually all changes in demand have no effect on savings and investment returns. It so happens that because of the Income Hypothesis, government spending has the multiplier above. Indeed, if one is being technical, government spending should have the same multiplier in or out of a liquidity trap. It would appear lower outside because of crowding out, and empirical studies and discussions don’t always attempt to disentangle these effects. Regardless of how we’re stating things, all the liquidity trap suggests is that, to a point, compensation shifts will have no effect on aggregate investment via the interest rate.

It’s not a coincidence, therefore, that I focused especially on government subsidies and the current account, though I didn’t give your borrowing proposal short shrift. The wage-productivity link would be an accounting necessity if not for those accounts to “sop up” the lost wages.

Corporate profits are superfluous here. They are repurposed into some kind of outlay, and thus show up elsewhere in the SAMs. It remains simply infeasible that, without the additional accounts to “sop up” lost wages, corporations would be repurposing it as anything but compensation. In light of that, your mechanism empirically does not hold.

R. A. Stark

April 10, 2014 at 8:29 PM

• The function of fiscal stimulus is to take funds that are idling in the hands of investors and give those funds to consumers, with the investors receiving bonds as compensation.

The function of a minimum wage hike is to take funds that are idling in the hands of investors and give those funds to consumers. There’s no compensation (though a wage hike could be accompanied by a tax cut).

Neither fiscal stimulus nor a minimum wage hike is expansionary when consumption is not the limiting factor on growth, but both are expansionary when it is.

Corporations are certainly putting the profits they’re sitting on somewhere (often into bonds), but it’s not somewhere that is getting the money to consumers so that consumption ceases to be a limiting factor on growth. The premise stimulus and proposals for a wage hike operate under is that there is a limiting factor on growth and that this limiting factor can only be alleviated if funds are redistributed throughout the economy to remedy it. Why would it matter whether those funds came from corporations in exchange for bonds followed by government spending as opposed to in the form of direct wages to workers (perhaps accompanied by a tax cut)? It’s the same idling money we’re moving around here. Unless you’re denying that there is any money that is idling? But if that’s the case, why does fiscal stimulus work in the first place? Government has to get the money it uses for stimulus from investors, either through bonds or taxes.

Benjamin Studebaker

April 12, 2014 at 8:57 PM

• What makes this response so frustrating is that I understand how the fiscal multiplier works—and how it is theoretically derived. It is done by positing (based on an extraordinarily sound empirical relationship) that marginal changes in consumption are proportional to marginal changes in output:

C=cY+k

where c is empirically about .75 and k is some constant that doesn’t matter terribly much to the present discussion. By plugging this into GDP (Y=C+I+G+NX) and rearranging, you can show that there is a theoretical multiplier of 4 on changes to non-consumption accounts of GDP. The fiscal number is related to this multiplier, but requires some additional behavioral assumptions and derivation.

The first relationship means that insofar as you are correct that your policy is a credit to Consumption, your policy will be a credit to Income and Production. (At a rate of about 133% of the new consumption, I should add!) It’s an even straighter shot than the fiscal multiplier, because the fiscal multiplier is in fact a consequence of this empirical multiplier.

The basis of my critique is not on the fiscal multiplier, which is irrelevant to the present discussion, nor to the original consumption multiplier. You’re correct that these things have the effect you describe. Rather, my critique is twofold. First, that these production accounts do not hook up neatly to the income accounts as you describe. Neither reiterating the unrelated relationship I just detailed nor asserting as an afterthought that you expect Compensation (W) to hook up a certain way makes it so. Second, that even if they do hook up as described, the balance of the empirical literature calls into question the very notion that real compensation is increased by a minimum wage hike.

The liquidity trap qualification only applies to a *short run* *crowding-out* argument, which I know better than to make. The “idle money” metaphor is not useless, and I’m not contesting it. But the model that underlies it is the LM-IS (or related) model. When there is a demand increase, demand for money shifts to goods, and there is an interest rate increase. As a consequence, there is a net decrease in investment, though it is not enough to offset the original demand increase. When interest rates are near zero, money in balance sheet savings (but not macroeconomic savings) outpaces money being spent on capital—that is, investment. You could say that money is idling on the books without much fuss from me. (Though, it’s important to remember, it is earning a higher rate of return than buying capital would.) The point being, as long as the question is whether or not your policy crowds out investment, the liquidity trap qualification matters.

I am not asserting that increases to total compensation crowd out investment, so there is no reason to parry this. The money is sitting idle in some sense, so the burden is to show that your policy moves it. (Although, I have some questions about a permanent wage hike as a means to combat a liquidity trap once that liquidity trap ends. If you do manage to hook this up on liquidity trap grounds, you’ll need to show that a permanent hike won’t be responsible for crowding out; that you had to use a liquidity trap rebuttal in first place would imply you can’t do that.)

The longer we have this discussion, the less time you seem to be spending on the income accounts in favor of the product accounts. The product accounts play a role here—see my post for how. But we agree on most of the key mechanisms. Where we disagree, and where you need to engage, is *how* precisely an increase in the wage-rate will move money into the consumption account. It’s not sufficient to show the money exists or that crediting other accounts would have the desired effect. The question is if a credit to compensation happens and if that translates to a GDP effect.

R. A. Stark

April 12, 2014 at 10:35 PM