After my previous post on US real wages, the original FEE article which prompted the initial reader question was brought to light for discussion. In short, this article suggests that the US middle class has actually done better than was suggested by multiple economic studies. Our reader was skeptical.
If nothing else, this should be a cautionary tale about the nature of economics- particularly macroeconomics, which dabbles in aggregated data and statistics. Economists can spin data, and a story to suit an agenda.
The reasoning in the FEE article goes something like this: The Pew article which the FEE article was criticizing takes brackets around the median (most commonly occurring) household income, defining low income as less than 67% of the median, and high income as greater than or equal to 200% of the median value, with the portion between 67% and 200% of the median defined as “middle income”. Based on this relative shift with respect to the median over the years, the Pew research paper concludes that the middle class is shrinking.
On the other hand the FEE article refutes this stance by pointing out that the income distribution flattens from the 1970s, and the median shifts right. This spreading out of the distribution (fattening of the tails) means that a lower percentage of families are classed as middle income based on the Pew article’s brackets method. Taking the actual household incomes shown in the distribution above they argue that the Middle Income bracket has shifted up.
While this might be true, I would like to raise a few interesting points which neither of the articles address. This took a little digging around in literature as my textbooks on monetary policy and the BLS documentation did not contain much information on this.
The BLS Inflation Measure Methodology Changes Over Time
Interestingly the BLS has changed their definition (and the fundamental underlying calculations) for the CPI on several occasions since the 1970s. The pre-1990s methodology relied on the Laspeyres Index. It is difficult to explain how this calculation functions in words but the mathematical formula is present in the link. It is, in essence, an index calculation which relies upon the disparity in price between a fixed basket of goods over two time periods.
The post-1990s methodology (switched during Bill Clinton’s presidency), claimed that the Laspeyres Index methodology overstated inflation. The argument put forth by Boskin and Alan Greenspan suggested that if consumer goods rose in price, consumers would switch to different substitute goods. Thus, the Laspeyres Index methodology did not account for this switch. Greenspan shifted from the previous arithmetic weighting system (Laspeyres) to a geometric weighting system.
Again during the Bush Era more measures were introduced. In short, critics have reported that there has been a tendency to understate the inflation rate in order to reduce social security benefits (the payments of which are adjusted for inflation.)
You can find more about this here.
Should the change in the reported statistics lead to lower reported value of inflation than is really present in the economy, and there is anecdotal evidence of this, based on the monetary aggregates, it is entirely plausible that both of the articles above are drawing incorrect conclusions. Afterall, even if you have the most perfect model but you feed junk data into it, you’ll draw junk conclusions. Note, some BLS employees have allegedly spoken out about this issue also.
What is Happening at the Far End of the Income Distribution?
Returning to the shifting FT distribution above, take a look at the 200K and greater bracket of wealth as it takes off like a rocket. What is happening here? Well, we touched on this briefly in my previous post where I mentioned that this is the equity (ownership) portion of the wage earners. This actually further validates the academic study that I referenced in the post.
Why is this occurring? Well there are a few reasons. Firstly, equities are a pretty good investment to protect wealth against inflation, but it’s also likely that the wealthy are more financially sophisticated than the average American. So it’s likely they have outperformed the lower brackets given the inflation over the years.
A second and more telling reason has a lot to do with the regressive nature of US capital gains taxes. Basically, if wealthy people keep cash in equities, they get rather sizeable tax breaks relative to if they were earning the cash as income. This once again benefits the equity (ownership) portion of the workforce.
A Little Experiment in Compound Interest
Assume that the average government reported rate of inflation was 2% between the years of 1990 and 2015. I will show how an under reporting of that figure corresponds to a reduction of purchasing power of a single US Dollar in a consumers pocket, over the years based on compound interest. For the sake of accuracy, we would need to assume that this coin was stored in the person’s pocket for 15 years. If it was in a bank account we’d need to calculate the real interest rate on the account.
Average Inflation Rate. Present Value (1990) Future Dollar Value (2015)
2.0 Percent 1 Dollar 60 cents
2.5 Percent 1 Dollar 53 cents
3.0 Percent 1 Dollar 47 cents
5.0 Percent 1 Dollar 28 cents
10.0 Percent 1 Dollar 7 cents
So in sum, what both papers differ on, is the methodology of the approach and while they are correct in their assertions based on the methodology, there is a possibility that they are both drawing incorrect conclusions based on skewed data from potentially poorly measured inflation.
While this might seem like conspiracy theory, I would say that based upon the monetary aggregates versus reported inflation, and based on the incentives introduced by the social security benefits, the thesis that inflation is understated is a plausible one. It is also interesting to note that the measure that social security payments are tied to now have their own category altogether since the Bush era.
This would also correspond to most Americans feeling that they are losing wealth at a time when the official statistics show otherwise.
Also, the findings of both studies are not incompatible with the academic paper posted in the previous article. In fact, the wealth distribution shown supports it.