In February, core PCE prices expanded 35bps, less than expected and below the average of the previous three months. This note explains (via Q&As) why the underlying pace is stronger than implied by a face-value reading of the MoM, why there is little signal in the downward surprise, and why we can expect a payback in the next two months. The reason is that the downward surprise was concentrated in the so called “non-market” items which tend to be noisy. In our view, the Fed staff follows a similar approach when analyzing incoming monthly data.
To keep in mind
Core PCE prices are split in “market-based” items and “other than market-based” items (also called “non-market prices”). Movements in “non-market prices” are noisy and carry little signal about future core PCE prices. As such, when looking at monthly readings, one should put more attention to “market-based” prices and largely discount any surprise in “non-market prices”.
What are non-market prices?
The BEA publishes two indexes: PCE prices excluding food and energy items (also called “core PCE prices”) and market-based PCE prices excluding food and energy items (also called “core PCE prices market-based). Core PCE prices market-based are derived from observed market transactions for which there are corresponding price measures. Put it simply: market-based items measure “everything you can buy with your credit card and for which there is an observed transaction”. On the other hand, non-market prices measure either opportunity costs or items purchased on behalf of households. Examples of the first category (opportunity costs) are “casino gambling” or some items in financial services, while an example of the second category (goods and services purchased on behalf of households) is medical insurance (which is typically purchased by the employer).
How are non-market prices measured?
Non-market prices are estimated in three ways:
- Where possible, some items are imputed using CPIs/PPIs/FX indexes as inputs (i.e. “casino gambling” is imputed using CPI all items as a proxy for the opportunity cost).
- Other items (especially labor-intensive sectors) are imputed using a cost approach. In this case, the price of the item is proxied using wage data (“Non-profit institutions serving households” is an example in this category).
- Finally, the remaining items are estimated by the BEA using ad-hoc models and the assumptions made by the BEA are unknown.
For this reason, the estimation of non-market prices is not immune from critiques and it is difficult to interpret month-on-month movements most of the times.
Are non-market prices relevant in core PCE prices?
Figure 1 shows the weight of core PCE non-market prices in core PCE prices. Non market prices accounted for about 11 percent of core PCE prices in 1992. Since then, their weight has steadily increased and they now account for more than 15 percent. Therefore, non-market prices are relevant for the behavior of core PCE prices.
Figure 1
How do non-market prices behave compared to market-based items?
Figure 2 shows the MoM at annual rate of core PCE prices market-based (the blue line), and core PCE prices other than market-based (the yellow line). Non-market prices are (much) more volatile than market-based items and show a higher mean. Over the 2000-2019 period, non-market prices added about 2 tents to the YoY of core PCE prices and their standard deviation of monthly readings is about 8 times higher than market-based items.
Figure 2
For this reason, monthly readings of non-market prices should be heavily discounted and one should take more signal from movements in market-based prices. On this point, Detmeister et al. (2014) clarifies that the Fed staff considers deviations of non-market prices from market-based prices as noise (once the difference in means is taken into account).
Note: In the Fed staff Phillips curve described in Detmeister et al. (2014), all variables are specified in deviations from market-based prices. In this way, movements in non-market prices away from market-based end up in the residual of the model. From page 2 of Detmeister et al (2014): “ε is a residual, which captures sources of inflation variation (for example, unusual movements in nonmarket core PCE prices) that are unrelated to the model’s other determinants of inflation.”
What happened in the February report?
In February, core PCE prices expanded 4.3% (MoM, a.r.), below the average of the previous months (about 6.0% MoM, a.r.). However, the slowdown in core PCE prices in February was entirely driven by non-market prices. Core PCE prices market-based increased 6.1% (MoM, a.r.), in line with the previous two months, while core PCE prices other than market-based posted a -5.1% (MoM, a.r.), the largest contraction since 2011 (excluding early March 2020 readings).
What explains the large drop in non-market prices in February?
As explained above, non-market prices are a collection of heterogenous items and it is quite difficult to model their behavior. However, large monthly swings in non-market prices can be driven by specific items and, in some cases, can be anticipated before the BEA publication. Figure 3 shows the MoM (at a.r.) of core PCE prices other than market-based (the yellow line) and PPI portfolio management (the red line) which is used by the BEA as an input for the non-market core PCE item “Portfolio management and investment advice services”.
The correlation between core PCE prices other than market-based and PPI portfolio management is elevated (corr 0.60) and large deviations from the mean in non-market prices appear to be driven by PPI portfolio management, at least most of the times.
Figure 3
What does PPI portfolio management capture and how is it estimated by the BLS?
According to the BLS documentation, price movements for PPI portfolio management are based on changes in the amount of revenue a mutual fund manager receives for providing investment advice. To track price movement for the index, data on management fees are collected. The management fee is most often based on a percentage of assets under management or a certain number of basis points (FEE = BP*FV, where FEE is the management fee, BP is basis points, and FV is fund value). Because FV (fund value) is highly correlated to the level of the US stock market, the PPI portfolio management is also highly correlated to US equities. Figure 4 shows the YoY of PPI portfolio management (the red line) and the S&P 500 index (the green line). The correlation between PPI portfolio management and the S&P 500 is 0.89.
Figure 4
What is the contribution of the stock market to consumers inflation?
We estimate that a 1 percent movement in the S&P 500 results in 10bps movement in non-market prices in 2 months (and in about 1.5 basis point movement in core PCE prices at the current weight of non-market prices). On average, in the pre-Covid period, non-market prices added 2 tenths to the YoY of core PCE prices. Out of these 2 tents, about half (1 tenth) was due to the contribution of PPI portfolio management.
What can we expect going forward?
In February, non-market prices subtracted 15bps from core PCE prices (market-based core PCE prices grew at a solid 50bps MoM, while core PCE prices grew 35bps MoM). Given the strong rebound of the stock market in March, we expect non-market prices to add (overall) 15bps to core PCE prices in the next two months (unless the stock market reverses back in April). Therefore, assuming that core PCE market-based prices will expand at an average of 35bps in the next two months, we expect core PCE prices to grow at about 42bps per month in March and April.