Bancreek introduces two new tools to help visualize U.S. Personal Consumption Expenditures (PCE)

But the downside of this approach is that bringing improved clarity to another complex database creates more questions than answers. The number of “why” questions grow each time we use our new PCE tools, especially in conjunction with our CPI-U tools, as they tell confusing and sometimes conflicting stories about the nature of U.S. inflation.

So, instead of taking the approach of answering the myriad “why” questions that inevitably fall out of this new data, we have focused on progressing our understanding of the “what” question. What do we mean by “answering the ‘what’ question?” We mean providing a means for the public to better collect and analyze data to make sure we are all homing in on the right questions before we start searching for their answers.

To bring this distinction to life, let’s talk about what’s going on with motor vehicle insurance inflation right now. It’s no secret that motor vehicle insurance is, in the words of one of our favorite podcasters Derek Thompson (of Plain English) “skyrocketing” – rising 22% year-over-year in March 2024.

Source: Bancreek Capital Advisors, LLC

But even this staggering number discounts the extent of the devastation motor vehicle insurance has had on CPI-U. We calculate that this single line item contributed 57 basis points of inflation to CPI-U last month. Mind you, overall inflation in March 2024 was 348 bps. In other words, one seemingly random category drove nearly 17% of all the inflation the average consumer is experiencing in this country today.

With such numbers being reported by the BLS, it’s not a surprise that the “why” floodgates have opened on this topic. There’s no shortage of theories on what could be going on, from better data on driving patterns driving up premiums to higher repair costs to an increase in accident frequency.

While there may be some truth to many of the theories floating around out there, we’d argue that from what we know, it’s premature to start asking why. Asking why based on this one data point may skip a step in the analytical process. Rather, we’re better served hunting for more context to see if there are any other clues that could steer us down a different path before we put our faith in this one number.

So, we returned to interrogating the data, this time bringing PCE data into the picture. Doing this revealed something fascinating – contrary to CPI-U data, PCE suggested that motor vehicle insurance was not skyrocketing. As shown below, February 2024 inflation (the latest month available at the time of writing) was 7.8%, which was within one standard deviation of its historical data and moderating from its peak in the low-teens late last year.

Source: Bancreek Capital Advisors, LLC

Notably, if you do the same inflation impact calculation for PCE, motor vehicle insurance added just four basis points to PCE in February 2024. This means we are left with a situation where we have one item that is simultaneously breaking inflation and largely irrelevant to inflation. In other words, the impact of motor vehicle insurance on inflation today is in the eye of beholder… or we should say, the eye of the measurer.

We kept going deeper into the data, venturing into the PPI data, and finding that PCE’s measurement of motor vehicle insurance is essentially identical to PPI’s “Private passenger auto insurance” category.

Source: Bancreek Capital Advisors, LLC

So, the BEA appears to be using the BLS’ PPI measurement of auto insurance rather than using that of CPI-U, which makes sense given that the BEA’s PCE is intended to be a more holistic view of expenses from all payers rather than only consumer expenses. This means that after all this data sleuthing, the true question we should be asking is what is the difference in auto insurance methodology between PPI and CPI-U? We’ve sure come a long way from our initial question based on CPI-U data alone.

Unfortunately, this is where we decided to pause our investigation. As we went through the factsheets for both CPI-U and PPI, we quickly realized that the methodologies themselves are both complex and nuanced, requiring a substantial commitment of time to understand in depth. We urge you to read these factsheets yourselves if you are interested in the details, but here is the key takeaway: they are quite different from one another, with an R^2 value of just 0.47, as illustrated in the following scatter plot:

Source: Bancreek Capital Advisors, LLC

So, we’ll close this tangent by wondering out loud if methodological differences in the surveying and measurement of one random item has become one of the key drivers of today’s inflation concerns – or lack thereof, depending which version of inflation you prefer.

But we digress. The point of going down this rabbit hole was not to try to find the answer to this pressing question (although that would have been convenient if we did). It was simply to illustrate how having a more comprehensive data toolbox can completely change the course of questioning and make us more productive with our time. This is our hope with these new visualizations. They are a few more puzzle pieces that can help us achieve better understanding of the similarities and differences between the two most notable inflation indexes in the United States.

To that end, today we are proud to place two new tools into Bancreek’s data visualization toolbox:

Bancreek Inflation (PCE) Visualizer

If our first PCE visualization tool looks familiar, it should! We designed it to be a near complete replica of Bancreek’s inaugural visualization – the Bancreek Inflation (CPI-U) Visualizer.

By way of quick tour of the visualization, at first glance you will see many bubbles of different sizes and colors. Each bubble is one PCE item, plucked from Table 2.4.4U of the Bureau of Economic Analysis’ (BEA) National Income and Products Accounts data tables. Just as we did for CPI-U, we have selected what we believe to be the most granular collection of items that together sums to total nominal Personal Consumption Expenditures. We put the tool through a rigorous validation process to ensure that all chosen items were mutually exclusive, ensuring that the items you see on this visualization will sum to overall PCE inflation each month.

The size of the bubble corresponds to the inflation impact the item had in the selected month. We calculate inflation impact by multiplying the current month’s inflation for the item by its weight in the prior year period. For example, in February 2024 Motor vehicle maintenance and repair inflated 6.70% year-over-year, while its weight in the prior year period (February 2023) was 1.28%. Multiplying these two numbers together arrives at an estimated inflation impact of 0.09%, or 9 basis points (bps). This means that out of the 2.45% (or 245 bps) overall PCE inflation in the month, 9 bps was driven by Motor vehicle maintenance and repair.

The color of the bubble corresponds only to the annual change in that month. So, for Motor vehicle maintenance and repair, the color represents the 6.70% year-over-year inflation in the month. This is why it shows up yellow, indicating that inflation was higher up on the diverging temperature color scheme we have used to display unweighted inflation for each item.

Just as is the case with Bancreek’s Inflation (CPI-U) Visualizer, we’ve provided the user with the ability to select historical months in the filter towards the bottom of the visualization. However, thanks to the pristine quality of the BEA’s databases, we’ve been able to go back further in history with PCE than we were able to do with CPI-U – our PCE visualizer goes all the way back to 1990.

Bancreek Detailed PCE Item Analyzer

The second visualization we are releasing today is the Bancreek Detailed PCE Item Analyzer. We were inspired to create this visualization by the fantastic quality and duration of the BEA’s National Income and Accounts databases, which allow for very long-term analysis of the detailed items comprising PCE. With so much data available in such an accessible manner, we decided to just create a simple tool to hopefully make it a bit easier to visualize the historical inflation of each detailed line item.

Note: We'd recommend interacting with the embedded viz below on a desktop/laptop or tablet. It will likely not fully display on a phone due to its tabular design.

This visualization is divided into two sections. The top section includes the full hierarchy of all the building blocks of PCE, presented in the same order given by the BEA in Table 2.4.4U. The only addition we have made is to format the names slightly differently to more easily see where they live within the hierarchy. Towards the right of this table, we present two data points for each item: annual change in inflation in the selected month and year, as well as weight of PCE in that period. You can use the filter at the top of the visualization to select any month back to the start of 1990. We’ve also added a second filter to the top of the visualization to help search for specific items of interest. If you use this filter, make sure to uncheck the “All” box at the top before you start typing free form text into the search box. Then check the boxes of any of the items of interest once they appear. This will filter the Item Detail list to only the items you have selected. For example, in the below chart, we typed in the word “education” and then checked all the items that contained this word to focus the list only on these items.

Source: Bancreek Capital Advisors, LLC

To update the chart in the bottom section, simply click on any item in the top chart, and the bottom chart will display the inflation and weight history for the selected item. For example, in the figure below, we have clicked on “Prescription drugs.” The chart shows both the history of this item’s annual inflation (green series on left axis) and its weight of PCE (blue series on right axis). So, we can quickly see that Prescription drugs has steadily grown to become a larger portion of PCE over the past 30+ years, growing from roughly 1% of PCE in January 1990 to 2.75% of PCE in February 2024. You can also see that Prescription drugs current inflation of 0.25% is well below its historical average of 3.37% (the green dotted line).

Source: Bancreek Capital Advisors, LLC

For more detail on usability and methodology of both new visualizations, please refer to our PCE visualization methodology post published in conjunction with this report. 

In our recent CPI-U analysis we have spent a considerable amount of time writing about the “Big 3.” As a reminder, the Big 3 consist of the following items:

The reason we started calling these items the “Big 3” is because they are currently dominating CPI-U, together comprising 79% of CPI-U in February 2024.

Source: Bancreek Capital Advisors, LLC

As we show below, this same analysis looks markedly different for PCE. These items together are less than half the weight as they are in CPI-U, and as a result, contributed far less to PCE inflation impact in the month.

Source: Bancreek Capital Advisors, LLC

It turns out that these three items only together added 91 bps to PCE inflation in February 2024. While that’s still notable, it hardly deserves the Big 3 moniker, in our view. Meanwhile, these same three items added 248 bps of inflation to CPI-U in February 2024, more than overall PCE inflation in the same month (which was reported at +245 bps)!

When drilling in a bit deeper, we found the OER and Rent item indices to be essentially identical in CPI-U and PCE, which means that the difference in inflation impact is completely driven by the difference in weights in the two measures.

Source: Bancreek Capital Advisors, LLC

Source: Bancreek Capital Advisors, LLC

Meanwhile, motor vehicle insurance is more complex, with both the weight and measurement methodology driving differences between its inflation impact. We already covered the difference in reported inflation numbers at length, so no need to beat that dead horse. But compounding the gap in inflation impact for this item is the difference in weight – Motor vehicle insurance carries a 2.51% weight in CPI-U, while Net motor vehicle and other transportation insurance carries a 0.55% weight in PCE. In other words, motor vehicle insurance carries more than four times the weight in CPI-U than it does in PCE! Put both differences together (price and weight) and that explains how this one item can be a 57 basis point gut punch to inflation in CPI-U, while adding just 4 basis points to PCE.

To be fair, there is inherent risk in hand picking a few items from PCE and CPI-U and using them to explain away the differences between CPI-U and PCE. In commenting on the differences between the two inflation measures, the BLS writes, “there is no one ‘smoking gun’ that explains the discrepancy between the indexes (CPI and PCE Deflator). Rather, the overall discrepancy is the result of the accumulation of a number of small effects.” In other words, the devil is in the details, which is exactly the reason we pushed these new visualizations out to shed light on these devilish details in a more intuitive and interactive way.

But to completely dismiss the stark contrast in the Big 3’s impact on these two inflation measures would deprive us of a key learning – namely that CPI-U and PCE are very different measures of U.S. inflation, especially right now. In fact, had we only seen the item weights and not known the data sources, we would have sworn the data was from two different countries. This starts to make sense when you consider that one (CPI-U) is intended to measure the consumer’s experience while the other (PCE) is intended to measure the country’s experience, and as we all know, there are lots of other payers besides the consumer in this country (e.g. state/federal governments and employers for medical care). But we suspect that folks still conflate the two inflation measures – reading into one release and thinking it will have any bearing on the other. It’s true that historically there has been a very strong correlation between overall CPI-U and PCE.

Source: Bancreek Capital Advisors, LLC

But don’t let this high-level chart fool you. As we have studied throughout this post, when we look under the hood, we can see how different these two measures truly are, and we think it’s just now (thanks to stubbornly high housing inflation and this odd auto insurance divergence) that these differences are being exposed. As such, it may be dangerous to assume that the near-term movements in CPI-U have much read through to the near-term movements PCE, and vice versa.

So, from all this work, the only generalized conclusion we are comfortable in making is that inflation is not equal to inflation. Rather, CPI-U and PCE are looking at the nature of U.S. inflation from different angles, oftentimes measuring it using different methodologies. So, it’s no wonder we are now getting very different answers on the direction and magnitude of inflation in the U.S.

But look on the bright side. You now have two new data tools to add to your toolbox that can help you dissect these two metrics, shedding more light on their similarities and differences. Happy researching, and please don’t hesitate to reach out with any suggestions on how we can make our data tools easier to use and/or more intuitive.