What Should GDP Measure? 7/29
Not production; not well-being; and yesterday's numbers mean nothing
This is not meant to be a post about yesterday’s GDP numbers, but I will give my take below. The TL;DR is that the reported small decline in GDP over the first two quarters has no significance. Zero. Nada.
Actually, this is a follow-up to last week’s post. Interestingly, that one received some favorable attention on Twitter. I worry that some of this was just some people’s reflex of “Peter Thiel bad, therefore Kling good.” That is not what I was aiming for.
In any case, I wish to talk about the economic issue, which I think is very deep and puts me outside of the mainstream of the profession. Let me start with a quote from Robert Solow, in his speech accepting the Nobel Prize.
One can not do macroeconomics without aggregative relationships; and at least for the moment there is no substitute for macroeconomics.
The “aggregative relationship” that Solow worked on is called the aggregate production function. Much of mainstream economics of the past 75 years has viewed the economy through the lens of this concept. I think this was a wrong turn.
Imagine one factory, producing a single form of output. The measurement of this factory’s annual output is straightforward: count the number of widgets produced.
We care a lot about labor productivity, which is output per worker-hour. If we assume that all workers are identical, then measuring the factory’s labor productivity is straightforward. Divide the number of widgets produced by the number of worker-hours spent producing them.
The aggregate production function is a fancy way of saying that we are going to treat the entire economy as if it were this one factory. From that perspective, the goal of measuring GDP is to measure the output as if the economy were one factory. I am questioning this interpretation of GDP.
Incidentally, when we calculate labor productivity, we are going to treat every worker as identical. That is another dubious approximation, but I will not dwell on it today.
Even though I do not like using GDP as a measure of output, I do think that GDP measures something that is useful. I see GDP as a measure of what Adam Smith called “the extent of the market.” It is a measure of all of the goods and services that are purchased at prices measured in monetary units.
I think of economic activity as specialization and trade. GDP can be a fair measure of economic activity so defined. Taking that perspective on GDP tells you what to include and what not to include. When you get explicit compensation in monetary units for providing something, that counts as GDP. When you do not get compensation, it doesn’t count as GDP. The official statistics generally follow that principle.
As much as we would like to measure total output of the economy, in order to be able to interpret economic performance through the lens of an aggregate production function, we cannot. Here is just one example:
The dinner that you buy in a restaurant is counted in GDP. The cooking that you do at home is not.
There is a lot of production that takes place outside of the market. But there is no way to measure the value of that production. What is the value of the work you do cooking at home?
We also would like GDP to measure how well the economy provides human satisfaction, or what economists call social welfare, but we cannot. Here is just one example:
The sex you pay for in a legal brothel can be included in GDP. The sex you do not pay for cannot.
Most people feel that they get more enjoyment out of sex if they do not have to pay for it. (Robin Hanson’s post helps explain why this is the case).
This example shows that being counted in GDP does not make an activity “better” or “right.” It is not a moral judgment. In last week’s post, I was not saying that paid-for child care is any sense “better” or “more right” than parent-provided child care. But I do think that for the purpose of measuring economic activity, it is correct to count paid-for child care and not to count parent-provided child care.
In fact, “total production” is not a measurable concept. The economy is not a GDP factory. The variety of what gets bought and sold changes constantly.
Neither is “total satisfaction” or “total well-being” a measurable concept. We do not have a reliable way to measure the satisfaction of any one individual, much less a way to take a weighted average of individual satisfaction levels.
I cannot think of a valid use-case for treating GDP as a measure of either production or satisfaction.
You do not need GDP to know that the U.S. is doing better than Mexico. There are plenty of other indicators that will tell you.
You do not need GDP to know that we are better off than we were fifty years ago. People don’t long for: Ford Pintos; Princess phones; television antennas; restaurant choices limited to American, Italian, or Chinese food; having to stop at a gas station to ask for directions; infirmity and death at what today would be considered a “young” age of 60 or 70.
But for more refined calculations, using GDP gives you too large a margin for error. How much better off we are than Mexico, or how much better off we are than in 1972? If you use GDP statistics to answer such questions, you will get a number. But I don’t believe that it is a meaningful or reliable number.
I think that the best use for GDP is as an indicator of the scale of the economy. How do you gauge the size of government debt or deficits? Compare them to GDP. How do you gauge the value of the stock market? Compare it to GDP.
The latest numbers mean nothing
People want to use GDP as a way to “score” economic policy outcomes. I think that is particularly unwise.
Real GDP in 2021 Q4 is reported as $19.806 trillion. In 2022 Q2 it is reported as $19.682 trillion. A decline of 0.6 percent is less than the margin of error in computing GDP in the first place.
Over that same time period, the index of total hours worked increased by 1 percent. If you take these numbers at face value and use the mainstream interpretation, then productivity in the economy fell by 1-1/2 percent. It is not hard to come up with factors that might cause a loss of efficiency—disruptions from the war in Ukraine, continued Covid phobia, confusion about prices in the context of inflation—but I think instead we should just ignore the apparent decline in productivity. It is too small to take seriously, relative to the conceptual and computational shortcomings of the data.
My way of looking at the data is that we have two indicators of economic activity that are giving opposing signals. I think that the hours-worked number is more reliable than the GDP number, so my bet is that economic activity actually increased between the end of last year and the end of last month.
I might add that inflation and employment statistics that appear during one Presidency usually do not reflect the policies of that President. Today’s economy is shaped by decisions made years ago. For example, most of the logs for the recent inflation bonfire were put in place during the Trump Presidency, in what I called “lockdown socialism.” And some of the kindling can be traced as far back as the end of the Bush Administration, with TARP, and the early days of the Obama Administration, with the ARRA “stimulus.” President Biden just happened to be in office when the bonfire was lit. As for his own economic policies, the bulk of the damage that they cause will show up years down the road.