GDP versus final sales

Bob Barbera tells me there is information in the data on "final sales to domestic purchasers." Final sales to domestic purchasers is GDP minus net exports and inventory investment. It measures demand for goods and services from US households, businesses and government, regardless of whether those goods and services are imported or domestically produced. GDP, by contrast, measures demand for US-produced goods and services, regardless of whether that demand is from foreigners or US residents. Here's a graph of quarterly growth rates of real GDP and final sales:

The GDP numbers show a weakening of growth, from 5% in 2009Q4 to 3.7% in 2010Q1 to 2.4% in 2010Q2 (the latter number likely to be revised further downward). But the final sales figures show the opposite: growth in demand accelerated from 0.2% in 2009Q4 to 1.3% in 2010Q1 to 4.1% in 2010Q2.

What does this mean? It means that the slowdown in GDP growth is due to a dramatic rise in imports - the BEA's report on 2010Q2 GDP says that import growth reduced GDP growth by 4 percent in the quarter. People - most importantly, businesses - are buying, but they're buying imported goods rather than domestically-produced goods. The result is little feedback to domestic employment. If I had to guess, I would guess that the problem is that weakness in the housing sector means lower than normal demand in sectors heavy on domestic production. Business investment in equipment and software has been leading growth the last few quarters, and much of this stuff is probably imported.

Some quick comments on the GDP report

The BEA reports that second quarter GDP growth was 2.4 percent, about what people were expecting but disappointing. At the same time, the BEA adjusted its estimate of first quarter growth way up to 3.7% (while adjusting downward its estimates of growth in 2009Q3 and 2009Q4).

There's lots of commentary out there on the report (here, here, here for starters), most of it highlighting the negative. Most negative is consumer spending, which crept up at an annual rate of 1.6 percent. That bad boy has to get into the 3 percent range for us to have a strong recovery, but it's unlikely to do so while employment growth is slow, consumers are heavily indebted, and the housing market is dead in the water.

But let's look at the good in the report. Business investment was very strong: non-residential fixed investment rose at a 17 percent annual rate, mostly due to investment in equipment and software. As has been widely reported, businesses are flush with cash, profits are high, and this month there has been a flood of new bond issuance. All of these things suggest continued strength in business investment in coming quarters.

Another thing that is good in this report is that the only major component that did not contribute to growth was imports. Imports rose by an astonishing 29 percent (annual rate), reducing GDP growth by four percentage points. That rate of increase is not sustainable. As imports return to their historic norm they will do less to dampen growth.

Finally, if we look at the last three quarters in its entirety, we see growth averaging 3.7 percent. That's not great, but it's not awful either. It would be great if we could have a couple of years of 5, 6, 7 percent growth like we did in the 1980s, but the economy isn't structured like that anymore and this is a different kind of recession. I'm guessing that 4-5 percent growth is about as good as we can realistically expect, and we're not that far off.

But back to the negative: the trend is down, and that's trouble. Even if business investment stays strong and imports stabilize, the fading effects of the fiscal stimulus will bring growth down in the second half of the year. There's no sign that consumers are going to go on a tear or the housing market is going to improve. So it's possible we could have 2 percent or lower growth in the next two quarters, as a lot of analysts are predicting. I'm more comfortable with the Fed's forecast of 3%+ growth however. Either way, the economy could use another dose of monetary and fiscal stimulus.

Austerity

A likely outcome of the debt negotiations is large cuts in government spending to reduce the debt burden. How would this affect the economy? Well, in fall 2010 the conservative government of the UK announced a program of austerity intended to reduce the government's debt load and restore business confidence so that businesses would start to hire and invest again (sort of what Republicans argue will happen if the US reduces spending). How's that worked out so far? See for yourself:



Caution: you never want to attribute every squiggle in the data to a single event. Lots of stuff contributes to movements in GDP. Still...

Blimey, that IS strong growth

Initially I wondered why everyone was responding so enthusiastically to reports that UK growth in 2010Q2 was 1.1 percent. In the US we're concerned that second quarter growth might be "only" 2 percent. As it happens, though, the UK statistical agencies report quarterly growth rates, not annualized quarterly rates as in the US. So the 1.1 percent growth is the equivalent of 4.4 percent in US reports. Brilliant!

GDP growth and ISM figures

The Institute for Supply Management reports that its purchasing managers index for manufacturing was 59.7% in May. Its non-manufacturing business activity index was 61.1%. Numbers above 50 indicate expansion in that sector. I ran a regression of quarterly GDP growth on quarterly average manufacturing PMI and non-manufacturing BAI and I get:

GDP growth = 0.58+0.09*(PMI-50)+0.28*(BAI-50)

Plugging in a guess of 60 in each index for 2010Q2 (which is now two-thirds over), I get 2010Q2 GDP growth = 4.2%. Let's guess productivity growth = 1.5% at an annual rate; that gives us 2.7% growth in employment (annualized) or 300,000 net new jobs per month (we had +290,000 in April). Sounds right to me.

GDP growth is still subdued

GDP growth in 2010Q1 was 3.2 percent according to the BEA's advance estimate. That's weaker than I thought it would be - I guess the consensus is smarter than I give them credit for. 3.2 percent is not strong enough to bring the unemployment rate down at a satisfactory pace. For that we need growth in the 4-5 percent range for the next year at least.

On the broader question of what type of recovery this is going to be, however, the data for Q1 confirm (for me at least) that we're on track for a traditional recovery rather than the jobless recovery that so many economists are worried about. For the first three quarters of this recovery, GDP has grown at a 3.6 percent annual pace. Compare that to the growth rates in the first three quarters following the 1990-91 and 2001 recessions: 2.0% and 2.3% respectively. The last two recoveries were jobless for a simple reason - GDP growth was too slow. We are on track, I think, for a recovery like that following the 1974-75 and 1981-82 recessions. Following those two recessions, GDP grew at a 4.5% and 6.1% rate respectively for the first two years of the recovery - I'd bet on a 1975-77 recovery rather than a 1982-84 recovery however.

Why the confidence? First, there's the fundamental statistical properties of GDP growth in the past few decades. If you run a regression of GDP growth on a constant, the output gap, and two lags of GDP growth, you find that typically deep recessions are followed by periods of strong growth: every one percentage point output gap adds about a third of a percentage point to annualized quarterly GDP growth. Our current output gap of about 5.9 percent therefore provides a powerful impetus to growth in the coming quarters: the model predicts a growth rate of 5.2 percent for the next four quarters. If we use the model to forecast the next four quarters of growth from a point three quarters into the recovery following the last two recessions, we get much lower forecasts: 3.6% for 1992 and 1.7 percent in 2002-03. Now the standard errors for these forecasts are very large, and in fact growth was somewhat higher than the model predicted in 1992 and somewhat lower in 2002-03 (4.2% and 1.7% respectively). So I'm not going to bet a lot of money on 5.2%. But the point is that strong growth coming out of a recession like this is normal, as is weak growth coming out of a recession like the last two. If you want to convince me that growth will be considerably slower this time around, you have to explain how the economy has changed. And no one has done that to my satisfaction.

The latest GDP report calls into question the most powerful argument that has been advanced for a slow recovery. Economists have been arguing that consumer spending is going to be sluggish during this recovery because households are heavily indebted and housing prices have fallen so much. But consumption spending rose 3.6 percent last quarter, following a 2.8 percent and 1.6 percent increase in the preceding quarters. As long as employment continues to pick up (and I think it will), consumption spending should be able to maintain the current pace for the foreseeable future. Business spending on equipment and software, traditionally a powerful cyclical indicator, was likewise strong last quarter, growing at a 13.4 percent annual pace. This indicates that business investment is not being restrained by problems in the banking sector, another of the headwinds that growth-pessimists focus on.

Weak spots last quarter were business investment in structures - a consequence of overbuilding in commercial real estate during the boom - and spending by state and local governments. There's not much we can do about investment in structures, but state and local government spending should be less of a drag as budgets improve along with recovery. Congress should long ago have provided more relief for state and local governments, and if I were in charge I'd press for this now.

The key to turning this into a self-sustaining recovery, of course, is growth in employment. Here again I think we can rely on statistical regularities for a powerful argument for employment growth in the coming quarter. A regression of employment growth on the previous three quarters' output growth suggests that in the second quarter employment should increase by 624,000 jobs, or just over 200,000 per month. If you think we're going to have considerably less employment growth than that, you have to explain how companies have been producing at the recent pace without hiring more workers. I don't think there's a convincing argument for that. It's more likely that we'll see stronger growth in employment than this - as Bob and I wrote in our Financial Times article a few months ago, companies probably overshot in reducing employment in 2008-09, and now will need to hire at a faster pace to get back to normal staffing levels. I'd bet on monthly employment gains of 250,000 or so for the next three months.

Anticipating Friday's GDP report

The consensus is that GDP grew at about a 3% annual pace in 2010Q1. I'm going to say 4.5%, based on everything Bob's been telling me, everything I read in the papers, and my sense that the consensus forecast is always too low in the early stages of a recovery (and too high in the early stages of a recession). These guys are less optimistic about Q1 but positively giddy about Q2.

Did I say put me down for 3 percent? I meant put me down for 1.8 percent.

Looks like the pessimists were right (again), GDP grew at a 1.8% rate in the first quarter. Notable weak spots were residential investment and government spending.

The following graph is from Calculated Risk. Note the contrast between the typical movement in residential investment following recessions and the movement following this recession. Several years ago Ed Leamer wrote a paper called "Housing IS the Business Cycle" and this graph demonstrates his point. Housing investment is a big component of typical recoveries and it has been notably, abysmally absent in this one. I don't think we can expect a strong recovery until housing recovers.



In light of the weakness in the economy it's appalling that Ben Bernanke did not use yesterday's news conference to announce a third round of quantitative easing (or, as I suggested a few days ago, a negative interest rate policy). It's appalling that the President and Congress are arguing about how fast to reduce government spending rather than - what was that word we kept hearing about in the Congressional campaigns last fall? Oh yeah, I remember - JOBS.

Thinking about GDP

Well who isn't? The BEA releases its report for 2011Q1 this Friday, and the forecasts aren't pretty. No one is predicting GDP growth of more than 3% or so for all of 2011 and some forecasters are going as low as 1.8% in the first quarter. Well, these people are looking at all the components of GDP based on recent data releases and probably have a pretty good idea. Nevertheless, my bird's eye view does not support such pessimism. Credit spreads are low, as shown yesterday. Also, the ISM manufacturing and nonmanufacturing surveys show a lot of strength. Even with a dip in March, the nonmanufacturing composite index averaged 58.8 in Q1 versus 55.9 in 2010Q4. Manufacturing has risen to 61.1 from 57.9. Hours worked rose at a 1.6% annual rate from 2010Q4 to 2011Q1. That's hardly blistering, but any decent rate of productivity growth gets us well above 2% GDP growth. So I'm going to go with, hmm, umm, let's see,... 2.8%. Heck, put me down for 3!

Oops

The BEA reports that GDP rose at only 2.8 percent in 2009Q3, not 3.5 percent as reported last month. We seem to be moving away from rather than towards my prediction of 4 percent. There does not appear to be a single culprit: consumption of all kinds was a bit weaker than the BEA thought earlier, residential and nonresidential investment were both weaker, imports took more away from growth than previously thought, inventory investment added less.

The new data helps explain last quarter's employment numbers: with GDP growth at 2.8 percent (pretty close to the historical average growth rate), we would not have expected any improvement in the employment situation. The rise in unemployment and the continued loss of jobs is not a mystery due to structural changes in the labor market, but simply a result of weak growth.

I'd be a fool not to start to question my faith in a modestly strong recovery at this point. The point I made awhile ago (really Bob's point) still holds however: every quarter that inventory disinvestment continues at its historically high pace (still estimated at -133 billion in Q3) creates the possibility of a bigger spike upward in inventory investment in future quarters. So it's possible that some of the 0.7 percent drop in growth due to this revision will be added to growth in a future quarter.

An alternative explanation

Paul Krugman cites a firewall-protected WSJ article offering another explanation for the recent rise in Treasury rates: the attack on QE by Republicans has raised questions in financial markets as to whether the Fed will actually go through with the plan. This is plausible, but it's very difficult to distinguish this possibility from the others.

Here's a way to think about the effect of QE on long-term interest rates. Long-term interest rates should be equal to the average of the expected future rates on short-term securities plus a term premium reflecting the degree to which long-term securities are imperfect substitutes for short-terms. The expected future short term rates, in turn, can be decomposed into expected future real rates plus expected future inflation.

QE has a direct effect on the term premium: think of a two-part process whereby the Fed first trades short-term securities for long-term securities, thus increasing demand for longs at the expense of shorts and driving up the price of longs relative to shorts, reducing the term premium. Then the Fed conducts conventional open market operations (purchases of shorts) to keep the interest rate at zero.

QE can affect expected future real interest rates by altering markets' beliefs about the health of the economy. A stronger economy means higher real rates, weaker economy means lower real rates.

QE can affect inflation expectations. Higher expected inflation means higher nominal rates, lower expected inflation means lower nominal rates.

Perhaps we can distinguish between the possibilities by looking at the spread between the 30-year and 10-year Treasury yields. The Fed's purchases under QE2 will be confined to securities with a maturity under 10 years. Let's see, if QE is effective we should see yields on 10-year securities fall relative to those of 30-year securities, so the 30-year - 10-year spread should rise. If QE is not expected to hold we should see this spread fall. Expectations of an improving economy and inflation should affect the 10-year and 30-year identically, so there should be no change in the spread.

Roll the tape:



Hmm, the spread rises following the Fed announcement of another round of QE (though this only makes sense if the $600 billion purchases the Fed announced was bigger than expected; I don't think it was). Then it starts falling a couple of days before the letter from Republicans and conservative economists criticizing the policy (but a few days after Sarah Palin's tweet). So if you fudge dates a little you can get support for the WSJ/Krugman view (how often do you see those two names together behind one view?!). But if you look at where the spread is now versus where the spread was before the QE announcement, and you remember that both 10-year and 30-year rates have risen since Nov. 5, it seems like support for the "economy is strengthening" view.

And then we learn that GDP grew at a 2.5% rate in the third quarter rather than the 2.0% increase initially reported. The increase in the estimate reflects better numbers for consumption spending, inventory investment and imports. So some more foundation for the economy-is-strengthening view.

GDP report

Another poor GDP report from the BEA. There's no sense in sugarcoating it, but some commentators are excessively bleak. Mark Thoma, for example, quotes Calculated Risk as saying that "without the boost in inventories, GDP would have been barely positive in Q3." This is, I believe, the fifth report in a row in which it has been noted that the growth we're seeing is illusory because it's inventory-driven. To the extent that inventory growth represents production of domestically-produced goods, inventory growth creates employment and income, which generates consumption, which along with the original inventory adjustment adds to GDP. There's nothing wrong with GDP growth that is driven by inventory growth rather than something else, except insofar as we think inventory growth is a one-time phenomenon. But in this case inventory growth seems to not be a one-time phenomenon - it's been contributing to growth for 5 straight quarters! On the other hand, to the extent that inventory growth represents purchases of imported goods, it's really not a net contributor to growth and removing it wouldn't affect GDP: every dollar we subtract from GDP because it's investment in inventory of imported goods is added back in when we subtract the same purchases from the import line. I think the report is grim enough - no need to pile on.

There's something strange about, say, Dean Baker saying that "When an economy gets out of a steep recession, it should be soaring, not just scraping into positive territory" and using the 1982-83 recovery as a reference point, after people of a similar viewpoint (and I'm betting Baker himself!) have been saying for several years now that we should expect jobless recoveries from now on because the economy has fundamentally changed since the 1980s. The recovery has certainly not been as strong as I had forecast last year. I had thought that we'd do substantially better than the recoveries from the 1990-91 and 2001 recessions, but I never thought we'd do as well as the recovery from the 1981-82 recession. In fact, we've done a little better than the previous two recessions:



So should we be surprised that the recovery has not been stronger? No, not really. Should we be disappointed and hope for better? Absolutely. But if history is a guide - if this recovery is more like the last two than the one before - then the fact that growth was below par in the first year of the recovery does not mean it will not accelerate in the second.

GDP is up at last!

The Bureau of Economic Analysis reports GDP growth was 3.5% in 2009Q3, considerably higher than the consensus forecasts. Lots of commentary here.

My take, which is largely informed by conversations with Bob Barbera, is that people are being far too pessimistic about the strength of the recovery. For example, here's Bruce Kasman and Jan Hatzius debating whether growth in 2010 is going to be slow at 2%, falling to 1.5% by year's end (Hatzius), or fast at 3.5% (Kasman). If growth is even as slow as 3.5%, that would be very surprising given how fast the economy recovered from the last two deep recessions: in the first five quarters of the recovery from the 1974-75 recession GDP growth averaged 5.4%; in the first five quarters of the recovery from the 1981-82 recession it averaged 6.0%. Forecasters are thinking we're going to have a recovery more like those of 1990-91 and 2000-01 (2.9% and 1.8% respectively), but I don't know why that would be the case. Yes, there are impediments to strong growth in the current recession that didn't exist in the 1974-75 and 1981-82 recession - notably, households have to work through a big debt overhang and the financial sector is still very weak. But there are also forces that should accelerate growth, among them extraordinarily stimulative monetary and fiscal policy. Once growth begins, I would anticipate that all of the accelerator effects that made the downturn so scary - low growth --> problems in housing sector --> problems in financial sector --> credit crunch --> low growth - will begin to work in the opposite direction. The financial sector that looks so weak today won't be looking weak at all a year from now. The threat of massive foreclosures and a crumbling commercial real estate market will recede as growth solidifies. I think it would be prudent to have another round of fiscal stimulus focusing on aid to states, but I don't see grounds for pessimism overall.

As for 2009Q3 GDP, Bob has an interesting take. He says (I'm paraphrasing here) that the inventory numbers the BEA computed are almost certainly too low. They have inventory investment at -$130 billion in Q3 versus -$160 billion in Q2. Bob says - and he's right, I've looked at the numbers - that the US economy has never had such a string of large inventory reductions as the BEA is reporting. And these numbers are essentially a guess anyway - at this point, BEA has the inventory numbers for July, partial numbers for August, and nothing for September.

A more realistic estimate of inventory investment is that rather than improving from -160 to -130 (a +$30 billion addition to GDP in Q3), the improvement was 2 to 3 times that (so inventory investment should be between -100 and -70). Eventually inventory investment has to be positive; this will take a few quarters, but the process has to move faster than the BEA is proposing.

The implications for our estimate of Q3 growth: the $30 billion improvement added 0.94 percent to growth in Q3. Doubling or tripling that gives us an additional 0.94-1.88 boost to growth. Now about a third of the higher inventories, if it occurred, would be due to higher imports, so you have to reduce it proportionately. This would give us a net impact of 0.63-1.25%. That is, Q3 GDP growth should have been between 4.1% and 4.75%.

If Bob's right about the inventory numbers, then when Q3 GDP is revised next month it will be revised substantially upward. If he's wrong about the inventory numbers for Q3, that just means that we're headed for a stronger inventory adjustment in the next quarter or two - we could see 4%+ growth in 2009Q4 and 2010Q1.

After that, it's kind of a race between the momentum of the recovery versus the headwinds in the form of state and local spending, financial stress, etc. I'm betting on the momentum of the recovery.