Why Friends Don’t Let Friends Calculate 18 Month Not-Seasonally Adjusted Growth Rates

Bruce Hall seems to think calculating 18 month inflation rates (either annualized or not) is just fine. It is just fine. As long as you don’t do it using not-seasonally-adjusted data. If you do that, you really should be clear. Illustrative example for CPI below.

Figure 1: 18 month annualized inflation for seasonally adjusted CPI-urban all consumers (blue), and for not-seasonally-adjusted CPI-urban all consumers (tan), calculated using log differenes. NBER defined peak-to-trough recession dates shaded gray. Source:  BLS via FRED, NBER, and author’s calculations.

My vision last time I recall the numbers is 20-400 (rather than 20-20) uncorrected, but even I can see the two series differ substantially at different times.

In case reader CoRev is going to accuse me of manipulating the data, let me note that I use FRED series CPIAUCSL for seasonally adjusted CPI, and CPIAUCNS for seasonally unadjusted. All other calculations are pretty straightforward looking at the formulas in the legend box — unless one is unacquainted with the use of natural logs, or distrusts them. (Here is Jim Hamilton’s post on the use of logs and log-differences, if you don’t trust Menzie Chinn either because of his name or worldviews.

I must say, 18 months is kinda odd choice. 1 month, 3 month, 12 month, 2 years, 5 years, ok. 1.5 years, well, one should have a reason.

 

8 thoughts on “Why Friends Don’t Let Friends Calculate 18 Month Not-Seasonally Adjusted Growth Rates

  1. pgl

    “btw, pgl insists on using 13.3%”

    No Brucie – I never endorsed the abuse of NSA data or your redefining a year as being 17 (not 18) months. Those mistakes all belonged to you.

    Or do you consider my pointing out what you did was odd as being an endorsement of your serial nonsense?

  2. pgl

    ‘Source: BLS via FRED’

    Aha! That’s Menzie’s problem. He is using BLS data. Shameful!

    Brucie boy in his first comment here bragged he provided a link to some high schooler’s website. Now it is true that the high schoolers who Brucie relied on claimed they had done to BLS for CPI data. But they forgot to tell Brucie their data was NSA. Brucie had no clue until I pointed this little fact out.

    But come on Menzie – no fair using BLS data. We all should rely on second hand sources from questionable characters. JohnH certainly does!

  3. pgl

    “one should have a reason.”

    A lot of people here have noted Bruce Hall’s reason for ginning up a big number – he is a partisan hack that does precisely what Kelly Anne Conway tells him to do.

  4. pgl

    Brucie finally wrote:

    Bruce Hall
    January 21, 2023 at 9:55 am
    Menzie, I repeat, I used the seasonally adjusted data from the Fed for the calculations.
    https://fred.stlouisfed.org/series/CPIAUCSL

    January 2021 = 262.2
    … rapid inflation
    June 2022 = 295.3 (rapid upward trend stops)
    July 2022 = 295.3
    … low inflation
    November 2022 = 298.3
    December 2022 = 298.1 (low increase since June)

    The discussion among the grownups for the past few weeks is has inflation slowed. Brucie fires off his nonsensical 13.3% as a reply to this question. But note that he is now saying we USED to have higher inflation but NOW inflation is a lot lower.

    Now had he said this from the beginning we could have saved a lot of time. But NOOO – Brucie was not allowed to say this. Why? He takes his orders from Kelly Anne Conway.

  5. pgl

    How bad get inflation get?

    https://www.msn.com/en-us/news/politics/michael-cohen-says-trump-would-sell-classified-info-for-can-of-tuna-and-maybe-already-has/ar-AA16B5gp?ocid=msedgdhp&pc=U531&cvid=0fea5c3129ad4af1ba1b12d82402f53e

    Michael Cohen Says Trump Would Sell Classified Info for ‘Can of Tuna’ and Maybe Already Has

    But wait – the price of tuna has gone up a lot at my grocery store. OK I switched to salmon but I can see Trump selling off the nuclear code in exchange for a good piece of Ahi Tuna!

  6. pgl

    Kevin Drum notes how the Tories have turned back the clock on better health care in the UK:

    https://jabberwocking.com/breaking-brits-slashed-spending-on-nhs-and-service-cratered/

    In 1997 Tony Blair and his Labour government took power and increased NHS funding growth to 6% per year (in real terms). In 2009, Labour lost power and Conservatives slashed NHS funding, approving real growth in its budget that averaged about 1.5% per year….When Conservatives slashed spending growth from 6% to ~1.5%, service levels dropped and patient satisfaction when down. In absolute terms, that growth rate of 1.5% from 2009-18 compares to a real growth rate of about 3% for Medicare in the US. By any measure, the NHS is pretty starved for funds.

    Keep this in mind when Kevin McCarthy wants to cut Federal spending here as he will assuredly take the Tory lead and go well beyond it.

  7. joseph

    I don’t see anything inherently wrong with non-seasonally adjusted numbers, as long as that is clearly stated. It depends on your purpose. If you want to know how much consumers actually spent over a period, you would use non-seasonally adjusted numbers. If you want to evaluate a trend in inflation over a period that is not 12 months, you should probably use seasonally adjusted numbers.

    Over a 12 month period, both should give similar results. That is why the BLS in its monthly inflation release cites seasonally adjusted inflation for their month over month but in the same report cite non-seasonally adjusted inflation for their year over year number. If you use an oddball number like 18 months, you are kind of mashing together 12 months seasonally adjusted with 6 months of non-seasonally adjusted.

    By the way, seasonal adjustment is an elaborate process. A separate seasonal adjustment is determined for every item in the index. For example gasoline is high in summer and food is high in winter. Then all of the seasonal adjustments are combined using individual weights for each item.

    1. pgl

      Well said – especially this:

      ‘For example gasoline is high in summer and food is high in winter. Then all of the seasonal adjustments are combined using individual weights for each item.’

      Now Bruce Hall told us over and over again that gasoline prices were both high and important to consumers. So I guess he had to know his use of NSA starting in January and ending in June would overstate the SA increase in prices.

Comments are closed.