The Job Guarantee and National Accounts Gaming

Share  Tweet
Kevin Baird / Flickr

As I noted in a prior post, the easiest way to understand a job guarantee within the context of existing welfare state taxonomy is as a basic unemployment allowance combined with a work requirement. This is also sometimes called “workfare.” What’s important about this classification is that it recognizes the essential nature of a job guarantee, which is that it is a welfare program for unemployed people, though one with rather harsh conditions attached.

I understand the rhetorical importance of denying this description of the program. But it seems like the opposition to this description goes beyond rhetoric. In reading some of the texts about the idea (especially Hyman Minsky’s), it really does seem like there is a bit of a fetishization of national accounting categories going on. That is, advocates seem convinced the policy is good in significant part because of the way the national accounts would score the various quantities involved, even though that scoring may not reflect reality in any meaningful sense.

Gross National Product

At various points in Minsky’s writings, he notes that the national accounts do not count transfer payments towards the country’s gross national product (GNP), but that they do count payments made to government workers towards the GNP. And this is true: the accounts assume that $30,000 paid to a teacher equates to $30,000 of output, but that $30,000 paid to a Social Security recipient equates to $0 of output. So, Minsky reasons, if you provide an unemployed person a “job” instead of an unemployment check, you can increase the GNP by an amount equal to the unemployment check without the government having to spend any more money.

A worker on unemployment insurance receives funds without making a current contribution to output. If the same worker received the same income on a work relief or Work Projects Administration (WPA)-type program, then he can be presumed to have made a contribution to GNP equal to his income. If the WPA output is useful and is sold, WPA-type relief for the unemployed is less inflationary than unemployment insurance. If the WPA output is useful even if it is not sold in a market, then it makes a contribution to the well-being of those who find the output useful, and, presumably, a tax or user’s fee offset to all or part of the WPA spending could be collected.

Minsky makes the same observation about replacing some Social Security payments with wages for a public job.

Because of the way matters are measured, the impact on GNP of a dollar spent to hire leaf rakers in the public parks is greater than a dollar given in welfare or unemployment benefits. In 1975 the government distributed some $80 billion in Social Security payments. If 50 percent of the amount paid on Social Security had been spent on wages for the aged in a variety of work programs, then GNP would have been higher by some $40 billion. It is clearly a normative economic and sociological question whether it is better for a country to provide income for the aged through jobs, either in private industry or in make-work projects, or to provide income through transfer payments.

Replacing unemployment checks and Social Security checks with “wages” for “jobs” does get you a GNP boost, but only through a kind of meaningless game. The goal of GNP is to try to estimate the total value-added of production in the country. But making a Social Security recipient rake leaves for their $30,000 annual income does not really add $30,000 of economic output to the country in any real sense. Juking GNP stats by pushing quantities out of the transfer account and into the employee compensation account is kind of funny and interesting but not really something that should motivate support of the idea on the merits.

Unemployment Rate

A similar game seems to be going on when advocates talk about achieving a 0 percent unemployment rate. As with GNP, it is true that if you tell the people at the Bureau of Labor Statistics to score everyone in the make-work jobs as employed, then you can get to 0 percent unemployment.

But the purpose of the unemployment rate, it seems to me, is to provide an indicator of labor utilization and labor slack. Unemployed people who are required as a condition of benefit receipt to do a trivial task are not really being utilized in a meaningful way. And, insofar as such workers are ready and able to take up work in the non-JG sector, they should also be counted in an indicator of labor slack. As L. Randall Wray notes, workers in the JG program “are not ‘lost’ as a reserve army of potential employees.” So why get excited that the unemployment statistics keepers might be forced to scored them otherwise?


Lastly we have the concept of earnings. The fetishization of this concept is more philosophically rooted than the other two, but still nonetheless takes the form of playing games with concepts in a way that sidesteps what seems to be their underlying meaning.

For example, Minsky writes:

Social justice rests on individual dignity and independence from both private and political power centers. Dignity and independence are best served by an economic order in which income is received either by right or through a fair exchange. Compensation for work performed should be the major source of income for all. Permanent dependence on expanding systems of transfer payments that have not been earned is demeaning to the recipient and destructive of the social fabric. Social justice and individual liberty demand interventions to create an economy of opportunity in which everyone, except the severely handicapped, earns his or her way through the exchange of income for work. Full employment is a social as well as an economic good.

Elsewhere Minsky says things like a job guarantee will reestablish the principle that “one must earn one’s keep.”

The problem here is that the concept of “earning one’s keep” is normatively rooted in the idea that someone’s compensation is related to their value-add. The appeal is supposed to be that they are makers not takers, meaning that they contribute to the national output as much as they take out of it.

But the JG does not actually ensure that people “earn their keep” in the sense that it makes sure they are adding value commensurate to their income. Rather, it sets a fixed income and pays that to people without any regard to what their individual contribution to the national output actually is. It is true, as Minsky says above in the GNP section, that, in the national accounts, it will be assumed all of the JG workers are being paid an income equivalent to their value-add, but we know from our own understanding of the program that this is not what is happening at all.

As L. Randall Wray notes, “those workers whose productivity is substantially above [the fixed JG wage] will find jobs in the private sector; those with lower productivity will find [the JG].” That is, the very design of the program ensures that the workers in it are not “earning their keep” in the sense that their productive output matches their pay.

That JG earnings are not related to each worker’s output does not matter to me of course. I think the whole idea that people’s income should be pegged to what they produce is stupid, kind of incoherent, and totally unworkable. Likewise, the idea that dignity and independence flows from “earning one’s keep” strikes me as clearly wrong and kind of reactionary. But clearly JG advocates do think there is something important about that idea. And yet their preferred policy only adheres to that idea if you start to believe the statistical quirks caused by reclassifying welfare income as earnings actually reflect something deeper than the limitations of national accounting.