4 Questions
Warning: The first question is tricky. But I’ve noticed that economic blog readers are very smart, and I have confidence that you will get it right.
1. Take a look at “Figure 1” in this Mark Thoma post. According to the BEA, did housing prices reach a peak near the end of 2006? If not, about when did they peak?
2. Use the answer to question 1 to figure out why I think BEA inflation figures are worthless. Hint; housing is nearly 40% of the core CPI. Fortunately, it is only 18% of the PCEPI.
3. Is there any other nominal aggregate that might give a more accurate reading of the timing and scale of nominal shocks hitting the US economy?
4. The ECB is proud of their recent success in keeping inflation low and stable. Does this imply they believe Europe was not adversely affected by the worldwide collapse in AD that began in August 2008?
I will provide my answers tomorrow night in an update. Those who aren’t longtime readers may be surprised by some of the answers.
Update 4/7/10:
See, I knew you guys were smart. Declan nailed it with the very comment.
A few comments:
Declan is right that the monthly changes over the last year would be below the 12 month changes. This is the “average/marginal” distinction that you may recall from drawing demand curves and marginal revenue curves. When the 12 month average is falling at a steady rate, the marginal monthly change is lower, and falling more rapidly. When there is a “kink” in the rate of decline, as in about May 2009, then the monthly numbers must have fallen discontinuously to a much lower level. It looks to me like the monthly numbers went negative in the second half of 2009″”so I’d guess housing prices peaked about that time.
2. So the BEA said housing prices were rising 2006-09, right through the biggest house price crash in American history. And now that house prices are actually rising, they show them falling. Not good for a product that is 40% of the core CPI. The basic problem is that flaws in BEA procedures make it look like prices respond with a long lag to monetary shocks. There actually is some lag due to some prices being sticky, but much less than the BEA data suggests.
3. NGDP.
4. Yes, for the following reasons:
a. The ECB is able to influence AD, not AS.
b. If they target inflation, presumably they believe that inflation is the appropriate indicator of whether AD is higher or lower than desired.
c. If they have hit their inflation target, then they presumably believe AD is at the appropriate level.
d. If they don’t believe AD is at the appropriate level, but they are fine with the current inflation rate, then they should not be targeting inflation.
In my view, answer “d” describes the actual situation in the Eurozone. For instance, the Eurozone has done fiscal stimulus, which makes no sense if inflation is on target.
Tags:
5. April 2010 at 18:42
Ok, I’ll bite (spoiler alert?)
1. Housing inflation has been > 0 for the whole period covered by the graph, so prices have been continuously rising – no “peak” in 2006 or anytime else. (although given it is a 12 month number and how it is heading down I would anticipate a peak this year or even late last year).
2. Every other bit of data says house prices have been falling.
3. The four-letter answer to everything – ****!
4. Hmm this is getting a bit philosophical/rhetorical. I guess the pragmatic answer is “yes”.
I suspect there is a bit more to 2. and 4. though . . .
5. April 2010 at 19:47
1. According to the BEA, house prices peaked in the begining of Q3 2008 (but they really peaked in April 2007 according to the OFHEO and May 2006 according to Case Shiller).
2. They are “worthless” because the timing of the drop in core PCE with and without housing is the same and does not at all agree with the peak in house prices from their own data (with housing it should have dropped much earlier).
3. Nominal Expenditure growth really takes a dive in the second half of 2008, consistent with the timing of the drop in inflation.4. Low and stable inflation in Europe is consistent with a fall in AD together with a leftward shift in SRAS. So, NO.
5. April 2010 at 20:21
According to the chart, housing prices *have not yet peaked*–they’re still going up.
5. April 2010 at 21:12
1) Figure one refer to yoy housing inflation. According to the graph it has yet to go negative. In fact the PCE housing and utilities index didn’t peak until the first quarter of 2009.
2) Other indicies of housing prices peaked much earlier (e.g. 2006 for Shiller’s historical real price index). But your hint also suggests that the weight of housing is off, that with proper weighting core inflation would have peaked earlier than it did with core PCE. In fact core yoy CPI peaked in September 2006 at 2.9% whereas core yoy PCE peaked at 2.6% in July 2008 nearly two years later.
3) Nominal GDP was growing at better than 6% yoy until the third quarter of 2006. In fact it fell sharply from 6.6% to 5.6% that very quarter.
4) Don’t get me started. The ECB has had to do some major gymnastic type back bending to redefine its concept of “price stability” lately. Prior to this the standard was based on the average of the three members with the lowest inflation rates. Now that there is deflation in the Baltic States and in Ireland the definition of price stability excludes states with deflation much to the chagrin of Lithuania who had complained only a few years ago about lack of transparency on the inflation standards at the ECB. I guess my answer is that the ECB has to congratulate itself on something given that Depression has taken hold in those countries.
5. April 2010 at 23:15
1 The chart shows house price inflation not house prices.
2 Beats me.
3 Ummm, let me guess, is it NGDP?
4 No, keeping inflation low and stable would be quite an achievement in the face of deflationary forces, but I don’t know if that is why they are patting their own backs.
6. April 2010 at 00:54
1. Um, looks like they haven’t peaked even.
“2. Use the answer to question 1 to figure out why I think BEA inflation figures are worthless. Hint; housing is nearly 40% of the core CPI. Fortunately, it is only 18% of the PCEPI.”
Housing mostly responds to interest rates and monetary effects. It should be in some aggregate, so we can see the monetary effects. However if the AD effects are from non monetary sources, then it won’t pick up on them.
“3. Is there any other nominal aggregate that might give a more accurate reading of the timing and scale of nominal shocks hitting the US economy?”
Housing is extremely important for monetary effects so it would be a good one by itself for them. If the shocks are not monetary in nature then maybe looking at anticipated demand by measuring PPI. NGDP is useful after the fact, but is measured far too slow to be of use for this in real time, (good for after the fact analysis). Movement and spread in various interest rates would be good for this, if they weren’t price fixed by the fed in the short run. Is there anything like NGDP only compounded weekly/monthly?
“4. The ECB is proud of their recent success in keeping inflation low and stable. Does this imply they believe Europe was not adversely affected by the worldwide collapse in AD that began in August 2008?”
I think a single currency union and an EU superstate are more important to them than short term economic well being. Also, the ECB keeping inflation very low will help strengthen the power and use of the currency.
6. April 2010 at 04:38
See, I knew you guys were smart. Declan nailed it with the very comment.
A few comments:
Declan is right that the monthly changes over the last year would be below the 12 month changes. This is the “average/marginal” distinction that you may recall from drawing demand curves and marginal revenue curves. When the average is falling at a steady rate, the 12 month average is lower, and falling more rapidly. When there is a “kink” in the rate of decline, as in about May 2009, then the monthly numbers must have fallen discontinuously to a much lower level. It looks to me like the monthly numbers went negative in the second half of 2009—so I’d guess housing prices peaked about that time.
2. So the BEA said housing prices were rising 2006-09, right through the biggest house price crash in American history. And now that house prices are actually rising, they show them falling. Not good for a product that is 40% of the core CPI. The basic problem is that flaws in BEA procedures make it look like prices respond with a long lag to monetary shocks. There actually is some lag due to some prices being sticky, but much less than the BEA data suggests.
3. NGDP.
4. Yes, for the following reasons:
a. The ECB is able to influence AD, not AS.
b. If they target inflation, presumably they believe that inflation is the appropriate indicator of whether AD is higher or lower than desired.
c. If they have hit their inflation target, then presumably they believe AD is at the appropriate level.
d. If they don’t believe AD is at the appropriate level, but they are fine with the current inflation rate, then they should not be targeting inflation.
In my view, answer “d” describes the actual situation in the Eurozone. For instance, some countries have done fiscal stimulus, which makes no sense if inflation is on target.
Marcus, Does the BEA provide separate data for “house prices” and “housing prices,” with the latter including both owner-occupied and rental units? If so, that might explain why they show house prices peaking in 2008, even as Thoma’s graph implies they peaked after mid-2009. Do you have any links for the data showing a 2008 peak?
The other answers also provided some of the key points I was looking for.
6. April 2010 at 05:14
– In addition to the 12 month YoY data issue, are you asking about real or nominal housing prices? If real, we’re looking for where the lines cross (or would cross, but for the YoY vs. MoM issue).
– Their inflation data is based on implied rental price (“affordability”) not the asset price. Rental prices change very slowly as prices are renegotiated every year, and are sticky, so only a modest number roll over every month. Even when renegotiated, we don’t often see rapid appreciation unless a renter leaves the unit. (the dynamic is not unlike wages) Plus we have the YoY comparison issue – so there’s a double lag.
In any case, the asset price and implied rental price are not even the same thing – one of the common definitions of a real estate bubble is when asset prices deviate too much from implied rental price (suggesting that people are buying the asset for investment/appreciation purposes rather than to provide housing consumption). This is why, for example, some people believe there’s a bubble in Shanghai property…
– This brings us back to the stock/flow issue, which is commensurate with the wealth/income issue. The issue being that consumption in the US (and, indeed, jobs as well) is so dependent on wealth (e.g. asset prices) that the Fed needs to protect asset prices or risk a collapse of AD – the Greenspan Put. However, they like to pretend they aren’t in the business of protecting asset prices; admitting this would be politically inconvenient – and would involve admitting some uncomfortable truths.
– I’m not sure your point about the marginal/average house price came across well. Or, at least, it didn’t come across well to me. Heh…
6. April 2010 at 07:31
Scott
I confused “housing” prices with “house” prices and was thinking about levels and not changes!
No, the BEA does not provide data for “house” prices.
The main component of “housing” in the CPI (BLS) is “shelter” composed of: Rent of Prim. Resid., Lodging Away from home, OER and Insurance.
“Shelter” inflation turned negative in January/10 but OER and Rent of Prim. Res. are still slightly (0.2% yoy) positive.
6. April 2010 at 08:37
Even allowing that the metric is measuring “market rents” instead of house prices, the figure implies that rents were continuing to increase during the contraction, and have only flattened out quite recently, which doesn’t seem to square with my experience or news reports. It seems more likely that they lagged house prices, but hit a plateau (so, that 0% growth rate should have been evident) near the start of the recession two years ago and began falling shortly thereafter. “Figure 1” doesn’t reflect that picture of reality.
I think if we were to try and build a new and improved “housing” component of inflation from scratch then it would occur to us that it’s not a straightforward exercise of popping in house-prices. Rents are closer to what we would want, but it’s not an entirely satisfactory proxy either – since the character and value of the average rented properties often differs meaningfully from the character and value of the average owned or mortgaged property in any community. You can’t just wash away the difference by counting square-footage or bedrooms.
I think a required cash-flow per occupied housing unit weighted by each type of house, location, and including the character of the possession, etc., would be better. In was probably impractical to do this until recently, but I would think with modern internet-age real-estate databases like Zillow it would now be a feasible and helpful improvement.
6. April 2010 at 14:13
Scott,
Tit for tat.
I have three questions. They’re easy.
1) Which EU member entirely avoided a recession following the “Global Financial Crisis.”
2) Which EU member with a felexible exchange rate depreciated its currency the most in the wake of the “Global Financial Crisis.”
3) What conclusion(s) can you draw from your answers to questions 1 and 2?
7. April 2010 at 04:37
Statsguy, I was looking for nominal prices.
I realize they use rental payments. They don’t even measure those correctly, for all sorts of reasons. In a weak economy many landlords give a first month’s free, or even first two months free option to new renters. This is equivalent to a price cut. And rental payments on existing contracts aren’t prices at all, they are installment payments on debt contracts. Saying rents are sticky is like saying coupon payments on 30 year bonds are sticky—yes, they are sticky, but they aren’t prices at all.
In any case, if you are trying to develop a price index that will be used for purposes of macroeconomic stabilization, you need the market price of capital goods like machines, factories and houses, not the rents on existing contracts. No one it today’s economy sells output for that price. The price of an existing apartment building impacts the incentive to construct new buildings, not the rent on old contracts.
You said:
“This is why, for example, some people believe there’s a bubble in Shanghai property…”
I agree that this is happening in Shanghai, but I would add that it is clearly rational. Incomes have been rising at double digit rates for decades, and are expected to continue rising very rapidly for decades to come. In an efficient market the future expected value of assets should impact it’s current value. If Shanghai property was priced according to current income, I’d pour my entire life savings into it. In 30 years Shanghai will be a very rich city, and the property there will be up closer to the levels you observe in Tokyo, NYC, London, etc.
By “marginal,” I meant the house price change over a single month, and by average I meant the change over the previous 12 months. If prices had been rising at 2% a month for 6 months, and then started falling at 1% a month, then after 12 months the 12 month average would still show a 0.5% gain, whereas the marginal increases would have been negative for 6 months.
If the Fed wants to reduce asset price shocks, they should stabilize NGDP.
Marcus, Thanks for the info. Note that even if the yoy is still positive. The most recent months may be slightly below the peak, albeit still above 12 months earlier.
Indy, You said:
“Even allowing that the metric is measuring “market rents” instead of house prices, the figure implies that rents were continuing to increase during the contraction, and have only flattened out quite recently, which doesn’t seem to square with my experience or news reports. It seems more likely that they lagged house prices, but hit a plateau (so, that 0% growth rate should have been evident) near the start of the recession two years ago and began falling shortly thereafter. “Figure 1″³ doesn’t reflect that picture of reality.”
I completely agree, see my answer to statsguy.
In my view the ideal index would ignore existing houses, and just look at the prices of new buildings and repairs on old houses. For purposes of macro stabilization, the only prices that matter are the prices of output that requires labor. Existing homes don’t require labor. In the end I just throw up my hands and go with NGDP.
Mark, You may think they are easy, but I don’t. Off the top of my head I thought Poland avoided recession and Britain devalued the most sharply. I’m guessing both answers are wrong. I will be interested in your answer. I think I know the implication, however, flexible exchange rates make it easier to stabilize AD.
7. April 2010 at 05:05
Scott,
You’re mostly right. The correct answer to #2 is Poland, however (based on depreciation since July 2008). Thus this provides additional support for your answer to #3.
7. April 2010 at 06:48
ssumner – “If the Fed wants to reduce asset price shocks, they should stabilize NGDP.”
can I ask, which of the following statements in this chain seem innacurate:
1) Wealth, or perceived wealth, is a main driver of AD.
2) The Fed’s instruments have a direct and immediate impact on nominal asset valuations, but not a direct and immediate impact on total wages (which is a function of wage rate and unemployment).
3) Thus, the mechanism of causation is NOT “Fed actions stabilize expected NGDP and therefore stabilize asset prices”, but rather, “Fed actions stabilize asset prices and therefore stabilize expected NGDP”?
Empirically, however, I have no idea how to separate the causation, unless someone has thought up a nice instrumental variable. To do this, someone would need to answer the question “How can Fed monetary action impacted expected NGDP without impacting asset prices?” At least some portion of the Fed’s impact on expected NGDP would need to operate independently of asset prices. But I have a limited imagination.
Indeed, from my simpleton point of view, it seems that “expected NGDP” and “expected nominal wealth” are pretty close to the same thing – the latter being the discounted present value of the former, but it might be more politically correct to talk about stabilizing NGDP. “Stabilizing wealth” seems to imply “making sure rich people don’t lose money”, and we all know the Fed doesn’t do that.
7. April 2010 at 08:53
Scott,
The BEA does track home prices via the price deflator for single family home production. Without further research, the peak quarter and pace of decline suggests they use OFHEO/FHFA.
This measure of housing prices doesn’t make it into the PCE deflator or any derivative of that index as used by Thoma (or the Fed for the purposes of gauging inflation), but it should make it into the overall GDP deflator. In any case, by 2007Q1 single family home production was less than 2.5% of GDP, and by 2009 it was less than 1.0% of GDP, so the impact of even severe declines in housing prices would be muted in the overall GDP deflator.
Anyway, using that index, single family home prices peaked at 108.211 in 2007Q1 and bottomed (maybe) at 95.243 in 2009Q3.
See line 20, table 5.3.4
http://www.bea.gov/national/nipaweb/TableView.asp?SelectedTable=144&ViewSeries=NO&Java=no&Request3Place=N&3Place=N&FromView=YES&Freq=Qtr&FirstYear=2002&LastYear=2009&3Place=N&Update=Update&JavaBox=no
7. April 2010 at 11:39
@Justin –
“In any case, by 2007Q1 single family home production was less than 2.5% of GDP, and by 2009 it was less than 1.0% of GDP, so the impact of even severe declines in housing prices would be muted in the overall GDP deflator.”
CalculatedRisk has covered this repeatedly. Most recently
http://www.calculatedriskblog.com/2010/04/feds-kocherlakota-on-economy.html
“Back in 2005, several analysts argued I was wrong that a housing bust would eventually take the economy into recession – they said residential investment was only 6% of the U.S. economy! They were wrong because they didn’t consider all the add on effects – and the impact of financial distress. Now residential investment is only 2.5 percent of GDP, and Kocherlakota is making the inverse faulty argument. During previous recoveries, housing played a critical role in job creation and consumer spending. It isn’t the size of the sector, but the contribution during the recovery that matters – and housing is usually the largest contributor to economic growth early in a recovery.”
7. April 2010 at 13:40
StatsGuy,
Just to clarify, what I was trying to say was that single family home construction was such a small slice of GDP that large amounts of deflation wouldn’t be easily noticed in the overall GDP deflator measure of inflation. This is despite the fact that the BEA correctly captured the fact that home prices were falling at a good clip for 2007-2009. Outside of that, I agree with Scott that the heavy emphasis on implicit rent in the core PCE deflator (which I believe is the Fed’s preferred inflation measure) creates serious flaws as a measure of inflation, as the implied rent (non-cash) can zig as home prices (with real cash implications) zags.
I wasn’t trying to argue that a rapid decline in home prices shouldn’t have/didn’t have a significant effect on overall economic performance, although I agree with Scott that the impact was fairly mild until NGDP fell off of a cliff.
7. April 2010 at 14:45
… I was citing CalculatedRisk in support of your position. His argument has repeatedly been that the small % of GDP attributed to housing (which is reflected in various indicators used by the Fed) understates its importance. Should have been more clear, sorry.
7. April 2010 at 15:34
Thanks for the clarification.
8. April 2010 at 02:16
4d) I suspect the ECB does not regard it as its job to manage AD. Hence EU governments’ need to use fiscal stimulus in place of monetary.
For example, from here:
Now whether this should be the ECB’s objective is a different question.
8. April 2010 at 05:23
Thanks Mark.
Statsguy, I would say the first statement is wrong. AD is determined by future expected changes in AD (NGDP). If NGDP expectations are stable, employment will be fairly stable, and vice versa. If wealth is stable, then employment will probably also be fairly stable, but NOT vice versa. In 1987 there was a big stock market crash, but not much change in NGDP expectations. AD was very stable. There’s the example that best discriminates between the two transmission mechanisms.
Justin, Thanks for the info. In my view the big problem occurred in late 2008, when all asset prices started falling. Not just single family homes but multi-families, office buildings, factories, machines, commodities, etc. This dramatically reduced the incentive to produce assets. And I think asset output fell much more sharply than consumer non-durables and services.
The data you found is very interesting, and is certainly worth a post. Note that during the famous sub-prime crisis the price of residential structures fell less than 1% (2007:Q1 to 2008:Q2). Then over the next 5 quarters, falling NGDP reduced residential prices by more than 7%.
As James Carville might have said:
“It’s the falling NGDP stupid.”
A similar pattern occurred with nonresidential structures, but even more pronounced. They actually rose during the early stages of the recession, but fell sharply a few months after NGDP started plunging.
My hunch is that structure prices are probably a bit more volatile that the BEA figures suggest, but less volatile than Case-Shiller suggest (they are biased toward sub-prime markets). The truth is somewhere in between.
Statsguy#2, Good point. As you know I have argued that the housing bust may have contributed to a mild recession after December 2007, but it takes falling NGDP to produce the sort of severe recession observed after August 2008.
Justin#2, I agree with your point that even if the housing data problem was fixed, and even if other asset prices were included, the GDP deflator would not have provided anywhere near as dramatic a signal as NGDP. Some prices really are sticky.
Leigh, You are probably right, but that just shows how bizarrely dysfunctional the EU has become. The ECB is the only institution in Europe capable of doing macro stimulus. The individual countries are hamstrung by their small size, and EU limits on budget deficits. The EU needs monetary stimulus much more than the US. At least in the US we have a federal fiscal authority. I don’t even think fiscal stimulus is effective here, but in Europe it’s completely hopeless.
I would add that “price stability” implies level targeting, and they are not doing level targeting, as far as I can tell.