M2.1: GDP measurement would improve if it focused on production and let MELI focus on well-being.

Kuznets originally developed GDP to measure production, but it soon came to be the primary measure of economic welfare, but these are two very different concepts.  GDP has served both purposes because economic production is correlated with economic welfare, but the tens of thousands of statisticians around the world that work on measuring GDP would be able to accomplish both goals better if they could produce two separate measurements, one that focuses on net production and another that focuses on wellbeing.

Part of the tension between measuring wellbeing and production is the difference between measuring stocks and flows.  Wellbeing is a measure of the flow of human experience in a particular year and cannot be stockpiled whereas production can be stockpiled and we should be trying to measure the net change in the stock of human material wealth rather than gross production.

These goals are in tension because it is possible to consume a big part of the stock of human wealth in a fun party, thereby simultaneously boosting the experienced flow of human wellbeing and reducing net production.  Statisticians measuring GDP are forced to choose between measuring the increased flow of consumption as an increase of GDP or the destruction of wealth as a decrease of GDP.  The official GDP agencies have tended to prioritize the former in many cases perhaps because GDP is primarily being used for normative concerns about experienced wellbeing at any particular time.  For example, it would be fairly easy to estimate depreciation and subtract it from production as Martin Weitzman suggested in 1976, but that would serve to focus on the stock of wealth rather than the flow of consumption and would take the focus of GDP even farther away from measuring the current flow of wellbeing.  Similar adjustments for environmental concerns could make GDP more accurate at measuring changes in the stock of human production, but  face the same objections.  If Qatar pumped oil out from under a river bank and sold it to finance a big party, it would certainly boost the current wellbeing of Qataris, but it shouldn’t all be counted as production any more than if Qatar pulled gold out of a bank account and sold it to finance a big party.  There is some production (or value added) that is generated by transporting the oil/gold out of the bank and selling it, but most of the official addition to GDP is not really production, it is spending down the national resources.  If we prioritize measuring current wellbeing, then it would make sense for GDP to include the full market value of the oil/gold, but if we really want to measure production, then we should only focus on the value added.

The basic tension stems from the fact that people can produce a different amount than they consume at any point in time because resources can be stockpiled.  In fact, nature has already stockpiled a vast amount of resources that we can deplete.  The history of Easter Island could be a cautionary tale about the importance of measuring true net production rather than gross production.  The islanders enjoyed higher wellbeing in the present by spending down some storable resources (trees) that would otherwise give higher wellbeing in the future, and when they had destroyed all the trees, their population collapsed.  GDP would have looked very high until the trees were gone and then collapsed and that would correlate fairly well with their economic welfare.  However, their true amount of production during the GDP boom was not nearly as large as GDP because GDP would measure the destruction of trees as if it were production. Similarly, warfare results in both the creation and destruction of capital, but GDP only measures the creation and so GDP artificially inflates the productivity of warfare.

The above examples show how GDP sometimes prioritizes the flow of wellbeing over the measurement of production, but in other choices, GDP prioritizes production over wellbeing.  For example, most of the components of GDP don’t directly enhance experienced wellbeing at all.  Only consumption spending directly affects immediate economic wellbeing. The formula for GDP is:

GDP = C + I + G + NX

C = “consumption” spending by households including expenditures on durable goods like automobiles and education.  This category of GDP is the most directly related to current economic wellbeing.

I = Investment spending by businesses on equipment, inventory, buildings, etc.

G = Government spending on both consumption and investment.

NX = Net eXports = exports minus imports.

Neither net exports nor investment spending (whether done privately or by government) has a direct impact on immediate wellbeing.  Part of the reason why they are often considered to be components of wellbeing is due to their short-run correlation with wellbeing during recessions.  After all, GDP was created as part of the effort to understand and combat the Great Depression and it is useful for that end.  An increase in investment or net exports is likely to indirectly boost wellbeing during a recession by increasing wage income and thereby boosting consumption spending and so it is easy to understand the historical motivation why Kuznets and later macroeconomists valued GDP for helping understand and combat the business cycle, but the real measure of wellbeing is consumption spending and an increase in investment and/or net exports merely affects wellbeing indirectly by boosting household income and thereby consumption during a recession.

There is considerable short-run correlation between consumption and the other three components of GDP within business cycles because all forms of production require workers and when any category of spending rises, that requires spending more on wages which increases household incomes and that increases consumption. Because of this, most spending on investment and net exports only has an indirect effect upon experienced wellbeing via their boost to consumption spending.  Investment and net exports should be considered intermediate production that isn’t counted in measurement of current economic wellbeing similar to the way that research and development spending is not counted in GDP because it is not considered part of our final production goal.  They are just part of an intermediate stage in achieving the goal that we really want to measure.

On the other hand, an increase in investment could also indirectly hurt current wellbeing by displacing current consumption spending if there is full employment.  Although this may seem counterintuitive, it is easy to see that there is little relationship between consumption and GDP in international data where consumption varies greatly as a percent of GDP.  The following table gives some selected examples using world bank data to demonstrate some of the range of possibilities:

2014 Household consumption (% of GDP)

Qatar 14%
Equitorial Guinea 17%
Macao, China 20%
Luxembourg 31%
Singapore 37%
China 37%
Ireland 44%
Germany 55%
Japan 60%
USA 68%
Egypt 82%
Afghanistan 108%
Liberia 134%

Although there is a strong relationship between investment and net export spending on the one hand and consumption spending on the other in short-run analysis of recessions, this cross-section data shows remarkably little relationship between consumption spending and the other two components of GDP.  Using GDP rather than consumption for measuring wellbeing creates a tremendous distortion in international comparisons.  Liberian households are doing better than their GDP statistic indicates because they got 134% of GDP and Qatari households are doing much worse than GDP would indicate when they only got to spend 14% of GDP.  Mean GDP in Qatar was $97k in 2014 which makes Qataris look like they are almost twice as well-off as Americans with mean GDP of only $54k, but if you actually see an average neighborhood in Qatar, it doesn’t look twice as wealthy as the average American neighborhood because mean GDP is a misleading statistic for comparing Qataris with Americans.

Nicholas Oulton is one of the rare defenders of GDP from its many critics.  On VOXEU he claimed that “GDP per capita is highly correlated with other factors that are important for welfare. In particular, it is positively correlated with life expectancy and negatively correlated with infant mortality and inequality.”  Then he demonstrated this with graphs of extremely convincing data:

mortality-vs-consumption

life-expectancy-vs-consumption

inequality-vs-consumption

 

This looks like a convincing correlation between GDP and the other indicators of welfare, but Oulton performed a bait and switch.  None of his graphs show any GDP data at all.  Instead he showed the correlation between one particular component of GDP, household consumption, and welfare.  This is because consumption is more important for welfare than the other components of GDP and so he left out everything else.

Oulton goes on to say that median income is better than mean income for measuring wellbeing when inequality changes which is true, but then he ironically dismisses median income largely by arguing that median income is equivalent to a measure of consumption (which is true **link**) and he implies that consumption is inferior to GDP because GDP includes investment.  This is strange given that earlier he had focused on the correlation between consumption and welfare indicators.  He probably chose to exclude investment because investment is poorly correlated with the welfare measures he selected.  This is a very bad argument for dismissing median consumption/income.

Because GDP tries to measure both wellbeing and production, it sometimes prioritizes measuring production and other times prioritizes wellbeing.  Due to the inherent tensions between the two, GDP serves both masters poorly.  Economists should recognize this basic tension between the economic wellbeing which is dependent upon the flow of consumption versus real production which should try to measure changes in the stock of resources.


Weitzman, M (1976) ‘On the Welfare Significance of National Product in a Dynamic Economy’, Quarterly Journal of Economics 90: 156-62.

Advertisements