Have you peeled back the categories measured by the 8% inflation? This is stupid for me to say but, for example... Used car prices factor into that 8% number. But if you aren't shopping for a used car... it doesn't really affect you, right?
Housing affects mostly everybody. Same with energy. I don't truly understand the weighting or everything that goes into the 8% CPI getting tossed around and I understand that it comes out to "on average as a whole, you as a consumer are most likely seeing a roughly 8% increase in cost" but I wonder if it is worth calling out that "actual personally perceived inflation" might be less (or more) than 8%
My rent hasn't gone up, I haven't bought a different (new or used) car. Gas is more expensive surely but I think that's an extra $40/mo for me or so. I think Chipotle bowls cost about $3 more now... Not exactly life changing?
CPI only shows 5%, due to the lag induced by their rent counting methodology. So inflation as measured by CPI is understated if anything, not overstated.
Using the same formula as was used in the 70s would produce double digit numbers.
>CPI only shows 5%, due to the lag induced by their rent counting methodology. So inflation as measured by CPI is understated if anything, not overstated.
>Using the same formula as was used in the 70s would produce double digit numbers.
Can you elaborate on this? When and what was the methodology changed?
On point one, CPI uses a combination of "owners equivalent rents" and more traditional rent measures.
Owner's equivalent rent is basically just a survey where they ask homeowners how much they could rent their house for. So survey participants understanding of market rents may lag.
But more importantly, when they survey renters, they ask them what they're currently paying. So if somebody is in a one year lease, and gets surveyed in month 11, they will give a rent figure that's almost a year old. On top of this, CPI includes rent controlled units, below market rent units etc. In some sense this is "correct" because it reflects what people are paying... But the whole value of CPI is to be a forecasting tool. Using lagging metrics is bad design imo.
Finally, they only survey 1/6 of the housing stock each month. So the whole sample is not updated every month.
To your second question, the biggest change is that CPI used to use home prices rather than the OER measure. That metric would show 15-20% rather than the 5% we get from their current formula. Given that the shelter component is the biggest weight in the CPI, it would shift the number up a few points.
>But the whole value of CPI is to be a forecasting tool
Forecasting tool for what? Prices? Everything about the CPI is about measuring price changes that already occurred, not to forecast future price rises. If the price of widget goes up 10% year after year, that's all CPI is going to report. It's not going to report what the price of widgets are 10 years from now. If you want inflation forecasts, you look at TIPS spreads or the price of various swaps.
Once you understand that, most of the measurement choices make sense.
>Owner's equivalent rent is basically just a survey where they ask homeowners how much they could rent their house for. So survey participants understanding of market rents may lag.
That's fine because they're essentially insulated from the housing market, so the price they pay is effectively fixed for decades.
>But more importantly, when they survey renters, they ask them what they're currently paying. So if somebody is in a one year lease, and gets surveyed in month 11, they will give a rent figure that's almost a year old.
>Finally, they only survey 1/6 of the housing stock each month. So the whole sample is not updated every month.
Again, also fine because the point of CPI is to measure the cost of living for americans, and the cost of living for americans is largely fixed months in the past (or for homeowners, decades in the past). That said, I do think only updating 1/6 of the housing stock is a bit shady because it basically applies the lagged measure twice.
You ask me about the methodological "issues" then immediately turn around to defend them... Kind of odd. If you really felt so strongly about this, wouldn't you have already been aware?
The CPI is the primary tool the Fed uses to set policy, policy whose effect has a multi month lag of its own. So by using backwards looking metrics, we severely impair the ability of the fed to set appropriate policy in a timely manner.
If you think the backwards looking metric is still useful, then the Fed should create a new metric based off current market rates, and use that instead. it's intellectually dishonest to defend using backwards looking metrics in the CPI methodology as a forecasting tool for Fed policy. The current methodology clearly masks the actual current market rates for rent.
Including rent controlled units tells us nothing of inflation, by definition. So why does the Fed consider these?
Example: it took a whole year for inflation to be acknowledged as a problem, and rents have barely shown up in it at this point. There are a few percentage points higher on CPI to come from rent alone (assuming other factors stay constant)
Using current market rates is not forecasting anything, it's telling you what rents are today. Using rents from a year ago is backwards looking, pretty obviously. It's a current snapshot of what people are paying, not what price levels are. Which is a fairly useless metric, as the the intent of Fed policy is to influence market pricing, and the biggest use case for the CPI is to provide datapoints to help forecast the path of inflation.
> You ask me about the methodological "issues" then immediately turn around to defend them... Kind of odd. If you really felt so strongly about this, wouldn't you have already been aware?
Because I want to understand your position before arguing against it, rather than imagine what your arguments are and putting them in your mouth.
>The CPI is the primary tool the Fed uses to set policy, policy whose effect has a multi month lag of its own. So by using backwards looking metrics, we severely impair the ability of the fed to set appropriate policy in a timely manner.
It'll be nice if we had a forward looking metric, but that changes nothing about what the CPI is. The BLS publishes the methodology and/or goals of the CPI, so the fact that it's not forward looking isn't some sort of secret.
>If you think the backwards looking metric is still useful, then the Fed should create a new metric based off current market rates, and use that instead.
AFAIK they use a combination of present data + expert predictions to base their decisions. Using "current price for rent/housing" CPI might make it forward looking for rent/housing, but it does nothing for other components (eg. energy/food), because those prices aren't locked in for consumers. If you actually want a forward looking metric, your best bet are financial instruments linked to CPI and/or prediction markets.
> it's intellectually dishonest to defend using backwards looking metrics in the CPI methodology as a forecasting tool for Fed policy. The current methodology clearly masks the actual current market rates for rent.
I don't get it, is the CPI supposed to be the end all be all metric for interest rate policy? I don't think that's a position that I expressed, nor is something the fed holds.
>Including rent controlled units tells us nothing of inflation, by definition. So why does the Fed consider these?
>Example: it took a whole year for inflation to be acknowledged as a problem, and rents have barely shown up in it at this point. There are a few percentage points higher on CPI to come from rent alone (assuming other factors stay constant)
>Using current market rates is not forecasting anything, it's telling you what rents are today. Using rents from a year ago is backwards looking, pretty obviously. It's a current snapshot of what people are paying, not what price levels are. Which is a fairly useless metric, as the the intent of Fed policy is to influence market pricing, and the biggest use case for the CPI is to provide datapoints to help forecast the path of inflation.
I think the problem here is that rent, unlike most things, have their prices locked in months/years in the past. This is unlike most other things in the CPI. You don't lock in your gas prices 6 months in the past, nor do you lock in your supermarket bill. While it's true that prices in the present will eventually be paid by someone in the future, they're also not reflective of what the average american is actually paying today. If you use current prices for some goods, plus current prices for rent (rather than whatever BLS is doing now), then the CPI becomes a weird mix of current + future prices. Imagine a commodities index that is composed of 5 year futures for crude, 3 year futures for wheat, and spot prices for natural gas. What would that even represent?
By that logic we should use car payments from cars bought 5 years ago instead of car prices today. Why doesn't CPI do this?
We should use mortgage payments from house bought 10 years ago, not estimates of current rents.
If I buy 100lbs of canned goods from Costco 10 years ago, we should use that pricing instead of current price of canned goods too.
If you are arguing that using backwards looking rent levels make sense, surely you must agree with these changes too, which are 100% logically consistent with that viewpoint? Otherwise you are just being intellectually dishonest
To say that CPI reflects prices paid over market rates is not even accurate because we don't measure fixed costs for most goods in the CPI, only current pricing. Rent is the only exception, where methodology is not aligned with current market pricing.
The Fed uses CPI as their primary tool for gauging inflation yes, among many other factors such as labor market tightness and so on. Core PCE specifically. Their entire inflation target is built around this as a metric, if you weren't aware.
Why you feel the need to defend it is beyond me. Clearly from a pure statistical sense, using backwards looking data to assist with forecasting is statistical nonsense. You can try to compensate for the flawed metric, with your own forecasting, but why not fix the metric to begin with?
But think whatever you like. I can't respect your view unless you agree that we should use lagging factors across the board to make measurement methodology consistent. Otherwise what is even driving your view? Bias?
What you're describing is a consumer expense index, not what I would think of as a consumer price index. And looking at locked in expenses from the past is largely useless from a monetary policy perspective.
> By that logic we should use car payments from cars bought 5 years ago instead of car prices today. Why doesn't CPI do this?
>If I buy 100lbs of canned goods from Costco 10 years ago, we should use that pricing instead of current price of canned goods too.
Well wikipedia says it's something that economists are "torn" on[1], so maybe they actually should be doing it for consistency reasons! Searching around it looks like the bank of canada[2] and the imf[3] considered doing just that.
Also, apparently the whole reason the adjustment was put in place was because academics complained that the CPI was too biased in the upwards direction[4]
>We should use mortgage payments from house bought 10 years ago, not estimates of current rents.
1. Given that mortgage payments are fixed and housing prices have went up in the past decade, this approach would probably underestimate compared to OER
2. it still doesn't solve the issue that houses (or more specifically land), is an asset, not something you consume (the "C" in CPI).
>If you are arguing that using backwards looking rent levels make sense, surely you must agree with these changes too, which are 100% logically consistent with that viewpoint? Otherwise you are just being intellectually dishonest
Bold of you to assume that I'd disagree with it ;)
>To say that CPI reflects prices paid over market rates is not even accurate because we don't measure fixed costs for most goods in the CPI, only current pricing. Rent is the only exception, where methodology is not aligned with current market pricing.
Of the other goods in the CPI, how much % are durable goods? Of those, how long do people typically own those goods for? For instance, I agree that in in theory it's worth factoring this in for smartphones, but they make up such a small part of people's spending, and the time span is so limited (~2-3 years?) that it's not worth factoring it. This is as opposed to a house that costs hundreds of thousands of dollars, and people own for decades. In other words, maybe the inconsistency is there because they only bothered to adjust the biggest factor?
>The Fed uses CPI as their primary tool for gauging inflation yes, among many other factors such as labor market tightness and so on. Core PCE specifically. Their entire inflation target is built around this as a metric, if you weren't aware.
No, you're missing the fact that they have experts interpreting the metrics. They're not just applying some rule like "if inflation > 3 then raise interest rates". That's why there was the whole "transitory inflation" thing a few months ago even though inflation was way above the target. Given that, unless you think the experts there are totally incompetent and don't have this factored in, I don't see how it's really an issue. Presumably it's baked into their models already.
>Clearly from a pure statistical sense, using backwards looking data to assist with forecasting is statistical nonsense.
I agree that the adjustments are basically a smoothing function that make the CPI more "backwards", but removing it doesn't magically make the CPI not backwards looking. It's backward looking by definition. It's recording chicken prices collected last month. If you want forecasts the CPI is not it. You'll have to get them yourself (ie. experts and/or markets). See also "transitory inflation" from last paragraph.
>But think whatever you like. I can't respect your view unless you agree that we should use lagging factors across the board to make measurement methodology consistent. Otherwise what is even driving your view? Bias?
>What you're describing is a consumer expense index, not what I would think of as a consumer price index. And looking at locked in expenses from the past is largely useless from a monetary policy perspective.
I'll have to concede that CPI literally says "price", so therefore technically speaking you're right that it should consists of price first and foremost. That said, you failed to answer my question from last comment. If you had a commodity price index consists of a random assortment of future prices (of varying lengths) plus spot prices, is that something that people want? It seems at least somewhat reasonable to adjust the prices from the index so that they're all for the same time period, even if that did mean it wasn't following the "real" prices.
Housing mostly only affects renters (~40% of the market) and is also not evenly distributed geographically. While rents in Miami might spike 40% in a year, that affects only (0.4 * 900K)/100M of households in the US.
Housing affects mostly everybody. Same with energy. I don't truly understand the weighting or everything that goes into the 8% CPI getting tossed around and I understand that it comes out to "on average as a whole, you as a consumer are most likely seeing a roughly 8% increase in cost" but I wonder if it is worth calling out that "actual personally perceived inflation" might be less (or more) than 8%
My rent hasn't gone up, I haven't bought a different (new or used) car. Gas is more expensive surely but I think that's an extra $40/mo for me or so. I think Chipotle bowls cost about $3 more now... Not exactly life changing?