I recently finished reading “GDP: A Brief But Affectionate History,” by Diane Coyle. The book’s key theme is that GDP is a synthetic measure of a nation’s productivity, an abstract idea, with many judgment calls in its construction. Measuring GDP is not like measuring the depth of an ocean or the speed of a train. In the case of GDP, the “object” being measured is just an idea, based on a foundation of judgment calls and assumptions. GDP is not something with an independent existence waiting to be discovered and measured. Changing the assumptions and judgment calls changes the resulting GDP, and changes have been made many times in the past. Here are some notes from the book:
Prior to the nineteenth century there generally wasn’t the same interest in measuring and tracking national income like there is today since changes for most countries year to year were minuscule. Back then the idea of long term economic growth wasn’t really a thing. If you wanted to expand your piece of the pie you took from someone else’s piece of pie (i.e. sacking and pillaging), and then later someone else came and took back some of your pie when you weren’t looking. The pie itself didn’t expand much year after year on average like it does today. The Industrial Revolution changed that. The idea that the world economy just keeps expanding at low single digit rates is now largely taken as a given since it’s happened for so many decades now, but it’s still a relatively new phenomenon compared to all of human history.
It’s called “Gross” Domestic Product because it takes into account investment spending but not depreciation, so it doesn’t net out spending to counteract wear and tear and obsolescence. If it did factor in depreciation it would be Net Domestic Product. “Domestic” means everything that occurs within a nation’s borders, as compared to Gross National Product, which is the economic activity worldwide owned by a particular nation.
You can measure GDP in three ways, all of them theoretically equivalent to the other. You can add up all the output of the economy, all the expenditures of the economy, or all of the incomes. The expenditure-based model is the most commonly used in the media and in textbooks, though many countries will calculate GDP with multiple approaches, and some will publish the statistical discrepancy (the U.S. and U.K. do). The expenditure-based model of GDP is defined as:
Consumption + Government Spending + Investment + Net Exports
Measuring GDP in the United States became formalized during the Great Depression and World War II years, when President Franklin Roosevelt wanted a more accurate measure on how the economy was doing. When creating the GDP metric, Roosevelt wanted government spending included. During the Great Depression personal consumption was down but government spending was way up. Previous calculations of national income didn’t include government spending. Prior to the early-mid twentieth century, governments were quite limited in scope and size relative to a nation’s overall economy so factoring it in was less important. Additionally, the thought at the time was that government spending could only exist through taxation, so government spending was not a net gain in a country’s output, but a transfer. Disregarding government spending during the Great Depression and ensuing World War II, however, would not count the vastly increased government spending while only counting the vastly reduced personal consumption, making it look like GDP was decreasing, and no president wants contracting economic figures on his watch. There was argument back and forth on whether or not to include government spending, but in the end Roosevelt won.
It’s interesting to look at how things would be if you didn’t include government expenditures in GDP. If you ignored government spending you would now no longer count any of the revenue Lockheed Martin, Boeing, General Dynamics, Raytheon, and any other defense contractor makes selling to the government. Money contractors make building and repairing roads and freeways wouldn’t be counted, nor would any of the salaries for anyone employed by the government. Before GDP was formalized in the Roosevelt era, government defense expenditures were actually subtracted from the national income estimate, with the remainder the estimate of production left for non-war purposes, with the thinking that war was a necessary but non-productive expense. This is a good example of one of the many judgement calls behind GDP. There’s no real right answer here, but starting in 1942 when GDP data was first published in the U.S., all government expenditures were included as a contributor to GDP, marking a change from the last two centuries worth of thinking that the economy was just the private sector. The pattern of growth from before World War II, through the war, and after would have looked very different if, as before, government non-defense expenditures were not included in GDP and defense expenditures were subtracted.
Soon afterward, other countries adopted the U.S. model of GDP which included government spending. Not adopting this model while everyone else did would make your numbers look quite weak in comparison! The U.K. was the first to follow the U.S., and in 1947 the U.N. published the first set of standards for calculating GDP, based on the U.S. system. These days the U.N.’s System of National Accounts (SNA) defines the standard on how to measure GDP, which all countries are supposed to follow. While the SNA defines the standard, adhering to it is strictly voluntary, the agency has no enforcement ability, and every country has its own means of implementing the standards, which are influenced by the nation’s interest and ability to keep detailed records, a country’s opinions about certain SNA standards, and of course, by cheating.
When measuring consumption, only final goods and services are counted to avoid double counting. For example, nails sold to a furniture manufacturer are not counted, but the furniture sold to the consumer is. That’s a clear-cut example, but things can often get fuzzier. The entire value of the iPhone is considered an import from China since only final goods are considered and not all the intermediate steps, even though only the relatively low-value final assembly occurs in China. Another judgment call example is business software. Software bought for business purposes was originally not counted at all, considered an intermediate purchase and not a final good or service. Later the rules were changed and software is now considered a capital investment, and so now when a business purchases Microsoft Office it is included in GDP. Of course all these rules are easier said than actually implemented. This is easier said than implemented. It’s easy for a statistical agency to count software purchases bought directly from the vendor by Fortune 500 companies. It’s much harder to count the thousands of small, single proprietor shops that may buy the software from Staples or Amazon.
Double counting is not corrected for when measuring government spending, so things like office supplies and trash hauling that a government pays for do get counted, and then double counted when the office supply store and trash hauling service book their revenue. This is because the allocation between intermediate and final public services for government is considered too difficult to clearly discern.
A judgment call was made not to include an estimate of value for unpaid work. if the value of all this unpaid work such as house cleaning, child rearing, and volunteering were added, it’s been estimated it would create a 10-20% boost to U.S. GDP, depending on how conservative or liberal you wanted to be on what you wanted to include and what value you wanted to put on it. Not counting unpaid work also creates weird situations such as if a person marries his/her housekeeper but the housekeeper keeps doing the housework afterward, this will be recorded as a drop in GDP. Not including the value of all this unpaid work is as big a judgment call as including government spending. These calls could have gone either way.
GDP only considers output, and doesn’t factor in the explosion of choice that’s occurred in many areas: compare the number of TV channels, breakfast cereals, soft drinks, types of shoes, etc. you can choose from today vs. in 1940. Also, since GDP only considers output, it doesn’t factor in the sustainability of that output: GDP may be rising nicely but the planet may be getting destroyed in the process.
Social Security, Medicare, and Medicaid payments are not included in the government spending figure since these are deemed transfers from one citizen to another. While all government spending could (and used to be) considered a transfer and therefore not a part of national income, strictly disregarding Social Security, Medicare, and Medicaid is the way things currently stand. The U.S. could instantly boost GDP by relabeling all payments currently deemed “government transfers” to “government expenditures.”
Services aren’t well represented (or at least directly represented) in GDP growth. For example, you can’t point anywhere in GDP growth and say this component is due to Google’s free search engine. Google’s search engine is represented in GDP by the electricity Google uses to run its servers, the electricity we use to go online, Google’s ad revenues, and the salaries Google pays, but there is no search engine “product” being sold and counted. The same goes for Wikipedia, Facebook, Twitter, YouTube, and even down to the countless free blogs out there.
In 1987 Italy increased its GDP by 20% overnight by including an estimate for illegal and grey market activities. In 2014 the other members of the European Union started including their own estimates of illegal activities into GDP. The average is about 15% of GDP. The U.K. is on the low end with 4%. Greece provides a very generous estimate of 25%, and one of the things Greece did when it tricked the EU into giving it more loans that were based on the size of Greece’s GDP was to boost the impossible-to-prove-or deny illegal activities estimate. While the SNA now recommends including illegal activities in its GDP standards, the U.S. doesn’t. If illegal activities were included it’s estimated that it would expand U.S. GDP by about 7%. Illegal drug sales alone are estimated at $111 billion annually in the U.S., theft from businesses another $109 billion, prostitution $10 billion, and illegal gambling $4 billion.
I learned about Benford’s Law for the first time in this book. Really interesting stuff. In counts of things, lower numbers appear much more frequently than higher numbers since you must always count through the lower numbers to get to the higher numbers. Greece was eventually caught cooking their GDP numbers, and the made-up GDP figures violated Benford’s Law, something forensic accountants later found. After reading about Benford’s Law it’s strange this wasn’t caught much sooner, but I imagine this was largely due to everyone in the EU wanting the numbers to be true. A Greek official who pressed for more true-to-life GDP statistics was charged with crimes against the state. For a primer on Benford’s Law and how to build an easy check for it in Excel (seriously, EU, you can build a check for it in Excel), look here: