According to one widely adopted class of economic models, raising the inflation rate would be one of the most helpful things that could happen to economies in the situation currently faced by Japan and the U.S. Here I describe some new research relevant for testing that theory.
With the short-term interest rate stuck near zero at the moment for the United States and Japan, an increase in expected inflation would mean a reduction in the expected real interest rate. According to a wide class of economic models, such a development should help boost total spending and help the economy to grow more quickly.
One extreme, if less widely accepted, implication of that claim is that even policies that reduce the overall productive potential of the economy might actually prove to be beneficial insofar as they result in higher inflation. Examples include increasing the monopoly power of firms and labor unions, raising payroll taxes, or even the more crazy-sounding policies adopted as part of the New Deal in the 1930s, such as killing baby pigs and destroying good fruit in hopes of getting higher prices for farmers. Although many economists would hesitate to embrace such proposals as constructive policy measures, their potential usefulness is a logical implication of a very popular class of economic models, as demonstrated for example in recent research by Gauti Eggertsson of the Federal Reserve Bank of New York.
New research by Johannes Wieland proposes some neat ways to test that theory. Johannes is a graduate student at the University of California at Berkeley who I’m delighted to report will be joining the faculty at UC San Diego this fall. Johannes notes that there are a number of natural experiments that have been run with which we could test Eggertsson’s hypothesis.
One example is the Japanese Great Earthquake of 2011. What happened when a powerful tsunami took out a lot of Japanese productive capabilities? It certainly raised expectations of inflation in Japan, as one can see in the graph below of the rates of inflation that market analysts were expecting for Japan both before and after the earthquake.
|
And was that helpful for Japan’s GDP? The clear answer is no.
|
Another example that Johannes considers is oil supply disruptions coming from events such as the civil unrest in Libya in 2011. Again, this proved to be a shock that increased expected inflation in places like the U.S.
|
And again, the actual behavior of U.S. GDP relative to what forecasters had been predicting prior to the crisis can hardly be described as a welcome development.
|
Johannes draws on a lot of other evidence, looking more broadly at other fluctuations in oil production and prices, experience across different countries, and some ingenious evidence based on the behavior of the inflation risk premium. His overall conclusion is that adverse supply shocks have undesirable effects for economies even when they are at the interest rate zero lower bound.
Johannes attributes this failure of many mainstream models to their simplistic specification of the role of interest rates. He proposes a model with richer financial frictions. When consumers borrow at a different interest rate from the federal government, it is no longer appropriate to talk about what a given development does to “the” real interest rate. Johannes finds that models with these kind of financial frictions generate the more sensible (and data-coherent) implication that adverse supply shocks are, well, adverse.
However, Johannes notes that when you modify this feature of popular models, some of the other, less controversial implications of the models are also called into question. For example, there is a broad consensus among economists that expansionary fiscal policy may be a particularly effective policy tool for an economy at the zero lower bound, in part because it’s another device that may succeed in raising inflationary expectations. His study concludes:
In any data-consistent model the rise in inflation expectations from higher marginal costs
cannot be expansionary, so that fiscal policy becomes much less effective. Consequently,
in the calibrated model with credit frictions, demand-side policies are up to 50% less effective
than in a standard new Keynesian model. This suggests that policy makers should be
cautious in expecting large positive outcomes from such policies at the ZLB.
For an encore, can Johannes add some realism to climate models?
Unfortunately for Johannes, he will now have to endure the onslaught from the New Keynesians. They will not permit a challenge to their near monopoly control of public policy.
Great work and stay strong Johannes!
Oil price increases and/or structural increases in dependence on energy imports (Japan post-tsunami) would seem like special cases of inflation increase, at least to this naif.
How about we raise the minimum wage by 50 cents every six months?
I don’t understand how you and Menzie can run the same blog. Does he read what you write? If he saw this article, steam will literally pour out of his ears.
The example of Japan is quite simplistic, no? The loss of output due to the earthquake probably more than compensated possible impacts of higher inflation expectatives, thus invalidating the analysis, or am I missing something?
“This suggests that policy makers should be cautious in expecting large positive outcomes from such policies at the ZLB.” – Hope for Johannes that he has some backing from Harvard-Mankiw the dear Mitt Romney fan.
As this sounds not political correct for UC San Diego (at least under chair Mrs. V. Ramey, as I remember she likes the Tayler/Hall/Durlauf Stanford club), says Johannes.
Like some of the posters above, I think the examples you take have some issues and could always be said to be ‘particular’/’specific’.
In general, I find it easier to argue that, inflation, in a world where salary expectations are flat/negative/highly volatile act as a tax on consumption and thus investment.
i.e. inflation is more likely to lead to stagflation than to real output growth.
http://theredbanker.blogspot.com/2013/06/on-scott-sumner-monetary-policy.html
It’s more theoretical than your practical examples but I still think it’s an issue MMers would have to address and cannot dismiss by ‘well, earthquakes, oil shocks and wars really do destroy supply capacity and that limit is real’…
I would rather not just assume the assumptions in both the article or your blog post. Rather, I would like to see the data on Japanese inflation expectations increasing post-earthquake and pre-Abenonmmics. The data and reports I had seen indicate that inflation expectations fell as deflation set into the economy, particularly as over-strength in the Yen strangled exports and investment in export oriented industry. http://www.bloomberg.com/news/2011-12-27/japan-factory-output-falls-on-global-slump.html
Again, this is the ultimate problem in economics, it is very hard to run natural experiments where all other factors remain neutral and differentiate between “correlation” and “causation.” It makes it very easy to cherry pick data to support a particular ideological hobby horse (which for some James Hamilton has done because he now has grudge against Paul Krugman for what he perceived to the mistreatment of his friends Rogoff and Reinhart). Apparently 60,000,000 unemployed and underemployed Americans, Europeans, and Japanese are just an abstraction to him, or as Stalin said: “A single death is tragedy, a million deaths are a statistic.”
Average inflation in the u.s. has doubled since 1972, from 2% to 4%. This has corresponded to a dramatic shift in wealth to the upper class. The wealthy are best able to deal with the Fed’s money manipulation.
Of course, this has been an era when the u.s. outperformed places like Europe. So, everybody is better off, I guess from inflationary trickle down. Unexpected inflation lowers real wages, and transfers wealth from long term lenders (banks etc) to long term borrowers (firms with long term debt, mortgages).
Some constant inflation might help when real wages are sticky downward, due to money illusion. This is somewhat dishonest government, but it does appear to work.
Jim: a quick question from a non-economist reader. You mention:
“Examples include increasing the monopoly power of firms”
That doesn’t seem to be a case of *overall* inflation, but rather a case of specific prices (albeit possibly a lot of specific prices). At least as I understand how inflation could help an economy constrained by high and sticky prices, the idea is that *all other prices* except the sticky ones go up, normalizing relative pricing relationships and allowing markets to clear.
The supply side shocks discussed here seem to be cases where disequilibrium of prices is increased, not decreased. So I’m not sure how to reconcile these two views of inflation and what it means for the basic inflation = good story.
Without addressing the loss of GDP that resulted from the damage of the Earthquake this work seem at best very incomplete.
That’s a neat paper.
It certainly doesn’t read as “don’t reduce real rates, don’t spend,” though. What it reads as strikes me to be, “don’t expect miracles, especially from inflation expectations.”
Those are extreme and absurd examples. How would anyone expect a country with a collapsing government and widespread civil unrest to have strong GDP growth? And did anyone expect Japan to have better growth in the immediate aftermath of the tsunami/nuclear disaster? What a ridiculous post.
The Japan data indicates a temporary spike in inflation and a temporary boost in output.
Somehow this is clear evidence that temporary supply side inflation wont help to boost output.
The Japan inflation data doesnt seem to match with anything I can find anywhere.
Eyeballing that graph it looks like strong growth for about 9-12 months post earthquake.
Strange.
But note, when they were destroying pigs in the 1930’s the market was saying that no one should be raising pigs.
So was destroying pigs really that bad a policy
when no one could profitably raise pigs?
“This suggests that policy makers should be cautious in expecting large positive outcomes from such policies at the ZLB.”
Fiscal policy that results in a large budget deficit, all else equal, will lead to the creation of new financial assets in the private sector by introducing treasury securities. Furthermore, if a bank acquires and holds the treasury security by swapping Fed reserves, deficits will result in greater bank deposit balances held by the private sector relative to the counterfactual had this policy not been taken.
Now all is not equal. Consider why the zero lower comes into existence. This occurs partially because the Fed is responding to a declining demand for credit at the given rate. At the zero lower bound, if the private sector is on net paying down debt to the banking system, this has the effect of reducing aggregate bank deposit balances held by the private sector.
There are two countervailing forces occurring simultaneously which makes the expansionary fiscal policy appear less effective. To truly understand the effects of the expansionary fiscal policy in the face of private sector delevering, we would need to compare the present outcomes to the counterfactual where the policy change had not been administered.
In terms of supply shocks potentially driving inflation higher, this is not certain. Yes, inflation expectations may rise temporarily, but a negative supply shock on balance could result in a reduction in private credit creation relative to the counterfactual where the shock had not occurred. Those affected by the supply shock may have less of a desire to borrow by taking on new loans while the banking system may be reluctant to extend the entity new bank deposits since their collateral has been impaired.
In essence, the supply shock impairs the endogenous money/credit creation which counters the inflationary effects that result from the reduced supply side.
Is the New Keynesian prediction that negative supply shocks are expansionary as flawed as the claim that fiscal contraction is expansionary in the short run. 😉
These models (New Keynesian), originally designed and estimated to match normal times, robustly make two predictions at the ZLB: First, demand-side policies are very stimulative and, second, negative supply shocks (e.g., higher payroll taxes) are expansionary.
Here is the New Keynesian channel through which a negative supply shock becomes expansionary.
In contrast, nominal interest rates remain unchanged when the central bank is constrained by the ZLB, so that higher inflation expectations reduce real interest rates and consumption today expands.
Johannes adds realism with a financial friction that reverses the standard result in the New Keynesian model.
Because a negative supply shock reduces profits and share values, the net worth of financial intermediaries falls, tightening their balance sheet constraints. In turn, banks contract loan supply, the borrowing spread rises, and borrowers reduce consumption such that negative supply shocks are contractionary at the ZLB.
Perhaps Johannes results can be applied to Obamacare and Obamaair, when each is viewed as a permanent adverse supply shock. Obama’s policies are currently (and/or are expected to) increase the marginal cost of hiring an employee due to greater administrative burden on all firms and an increase in employer provided health care for firms that currently do not offer a plan , (excepting firms that pay the mandated tax penalty and dump workers into the exchange), while Obamaair raises the marginal cost of energy. Following Johannes model – both policies will reduce profits and share values from what they would have been otherwise, borrowing costs will rise causing consumption, output and employment to be lower than what they otherwise would have been in the absence of Obamacare and Obamaair.
Interesting post; I’ll have to pull down soon-to-be-Dr. Johannes’ papers.
Intuitively, I’ve never understood the argument that an increase in inflation is stimulative. Yes, new inflation reduces real prices and interest, but only of previously signed contracts—and those contracts are already part of the economy.
But new inflation likely INCREASES the real prices of NEW contracts: As behavioral economics has shown, market actors tend to be overly pessimistic about bad news. They will thus likely only agree to contracts with high nominal prices, in the hope of protecting real prices–and those nominal prices will likely translate to an INCREASE in real prices. That, obviously, is not a stimulatory response, but rather the opposite.
I don’t believe the premise or the execution presented is valid. On the premise, I would be shocked if anyone could present me with evidence that inflation expectations at the producer and consumer level (not forecaster) can be significantly altered by a single event (perhaps a major war would qualify, but I suspect the flight and destruction of capital and labor would make it a tough study). On the execution, as I alluded to previously, I don’t believe “Consensus Economic Forecasts” constitute a valid measure of inflation expectations.
“demand-side policies are up to 50% less effective than in a standard new Keynesian model. This suggests that policy makers should be cautious in expecting large positive outcomes from such policies at the ZLB.”
So they’re still positive, just smaller? Rather than expecting large positive outcomes, we should expect medium or small positive outcomes.
re: Japan. The Japan earthquake happened in March of 2011. Power outages left some manufacturing off-line — read direct hit to GDP — and the whole of Kanto was gripped by fear of black rain. What’s more, Japan’s economy appeared to bounce back over the next four quarters… On one hand, it looks like a poor example, on the other, looking at the data it’s not clear GDP growth wasn’t boosted in late 2011/2012.
Also, I suspect the Eggertson results don’t necessarily apply to inflation in imported inputs, since increased spending on gas could also crowd out domestic demand.
I don’t think the Eggertsson thesis is that supply shocks at the ZLB are expansionary. In fact, Eggertsson is quite explicit that this is NOT his argument. Rather, his argument is that three conditions must be present as necessary (but not necessarily sufficient) conditions for negative AS shocks to have an expansionary effect. The first condition is that the nominal interest has to be at the ZLB. Now Japan and the US under the Great Recession clearly meet that condition. But the second condition is that there has to have been some exogenous shock that excites deflationary expectations…not low levels of inflationary expectations and not disinflation, but disinflation. And Eggertsson also says the deflation must be “extreme.” That describes 1929-1933, but not Japan and not the US in the Great Recession. Third, the adverse supply shock must be transient with a known endpoint. This isn’t a small off-the-cuff condition. Eggertsson devotes several pages explaining why this was important. Indeed, Eggertsson says that if the negative supply shock is persistent, then it will be contractionary. So I think there is a real risk of Eggertsson’s paper being misrepresented here. That doesn’t invalidate Wieland’s thesis absent Eggertsson’s second and third conditions, but it also doesn’t invalidate Eggertsson’s argument, which was about the New Deal era only.
In some of Eggertsson’s other papers he emphasizes that it’s also important that the public’s expectations of future inflation paths accurately match actual inflation. Again, it’s not clear that Wieland’s paper deals with this. Although the first chart does not show Japan’s CPI for 2012, the actual value was 0.1%, which was a drop (i.e., disinflation) over 2011 levels. Presumably the Japanese were updating their inflation expectations throughout 2012, and updating them downward. Given that, it’s not clear that Wieland has actually demonstrated what was claimed. I think I could just as easily argue that the disinflation in 2012 led to lower GDP growth at the end of the year. Inflationary expectations may have ticked up slightly immediately after the earthquake (as did GDP!), but it’s equally clear that disinflation had set in by 2012 and not surprisingly GDP fell off by late 2012.
I don’t recall the Libyan oil supply shock being all that big of a deal. Why not the far more serious concerns over a war with Iran, which was the subject of at least one Econbrowser post? Or why not the ongoing debt limit hostage dramas in 2010 and 2011?
Marco, fladem, and others: I disagree. The tsunami reduced the physical capacity of Japan to produce output. The core Keynesian claim is that output is restricted not by the ability to produce but instead by adequacy of demand for what gets produced, and that Japan had substantial excess capacity. That’s why the models can claim that by reducing the capacity to produce, inflation expectations can be raised, demand will increase, and output would go up, not down. Of course, when you start thinking through what actually happened in Japan in detail, you are forced to acknowledge that there was no excess capacity in some key sectors, namely electricity production, while there may have been excess capacity in others, e.g., automobile production conditional on having access to adequate energy inputs. I agree very much that these sectoral issues are quite central to what goes on– indeed, this is my core belief. But in this respect I am taking a view of the world (and implicitly, you are as well) that is quite different from the New Keynesian models being discussed here.
Robert Bell: Continuing on the theme of my response to Marco, et. al., I agree very much that these real-world supply shocks are in fact relative price shocks that affect some sectors differently from others. One could in fact argue that the same is true of monetary policy, boosting housing for example relative to other sectors. Again I think this is a very important detail that is overlooked by many macroeconomists and is certainly left out of the models being discussed in this post.
mikev: Please note that the third and fourth graphs refer to U.S. inflation and output, not Libyan inflation and output. The question is whether the Libyan disturbances, by raising the U.S. inflation rate, might have been helpful for the U.S. economy.
jason11: Please note that the graph starts in 2010. The growth in Japan you refer to came before the earthquake, not after.
wcw and Nylund: I agree. A smaller multiplier is not the same thing as a zero multiplier. However, an important question of interest is whether the multiplier is bigger than one.
A political decision was taken to turn off 50-odd nukes in March of 2011:
Japan
Year Ended Year Ended
March 2012 March 2011
CONSUMPTION
Fuel oil 11,822,365 6,297,685
Crude oil 11,572,589 4,759,614
LNG 52,885,437 41,740,733
Coal 49,293,267 51,025,604
From 2003 to 2010, the majority of U.S. dairy farmers voluntarily subscribed to contribute (proportionally to their production) to a “herd retirement” program to slaughter entire herds of productive dairy cattle, eventually slaughtering roughly 500K head.
http://www.agecon.purdue.edu/undergrad/counseling/pdf_files/CarissaMcCayThesis.pdf
Inflation may not be helpful at restarting GDP growth, but it is extremely helpful at devaluing the debt of governments.
Check out US debt/GDP and the CPI in the years following WWII. Inflation was smokin!
Think we have a different plan for our debt this time? I doubt it.
”
strong GDP growth? And did anyone expect Japan to have
”
I postponed my Japanese origin purchases for one year. Nobody wants to buy something that is still glowing. But seriously folks, you can’t glean any reliable data from a totally manipulated captured market. Man, there is stuff going on between big banks and big guvvment that would make any economist puke. Worse in Japan?
Believe
it
!
@JDH — The point is that the Tsunami was also a demand-side shock. Check out japanese tourism to Hawaii after the Tsunami: http://www.hawaiitourismauthority.org/research/reports/historical-visitor-statistics/.
In case you don’t want to click through the link, Japanese tourism dropped like a rock after the Tsunami and then later rebounded.
Do you really think Japanese were vacationing in Hokkaido instead? The Tsunami had a big impact on the Japanese psyche, if they canceled their vacations (counts in demand), how do you know they didn’t also delay large purchases?
Please, pray tell your readers where in Eggertson it says declines in consumption triggered by fear of nuclear winter (totally overblown, but consider the place) lead to GDP increases.
Even so, Japan’s GDP growth after the initial decline does look very impressive by GDP standards. Tsunami rebuilding/broken windows?
I do agree that the counterintuitive results in Eggertson shouldn’t be applied to the present. For one, in Eggertson (I believe) there’s one interest rate, and generally one interest rate in liquidity trap models. Presently, there are many interest rates in the economy, and not all of them are zero => capacity for the Fed to do something. E.g., my student loan interest rates are definitely not zero. Thus supply-side shocks will impact the Fed’s behavior, and we haven’t contradicted the basic logic of Eggertson, since it’s not actually true that all interest rates are zero. It’s not even true that the federal funds rate is zero, although it’s certainly close… It’s also not true that the Federal Funds rate couldn’t be negative…
Jim has already responded to a number of your comments, but let me follow up on a few more.
@ benamery21 and others: Some of you have commented that the episodes I analyze are very special cases. If you look at the paper, I consider three different kinds of evidence: First, event studies (Japan, Libya) where the timing of the supply shock is very clear. Second, oil shocks from a VAR, which are harder to identify but give us more data at monthly frequency. Third, I use inflation risk premia to determine if financial markets consider inflation surprises good news. In all these cases I find that negative supply shocks have contractionary effects. Thus, this result is quite general but it cannot be explained within the confines of the standard New Keynesian model. And this result is important for how we think about other issues, such as the size of the fiscal multiplier.
@ Jason11 and slugbait2: The inflation data and the forecasts use 2005 CPI weights. The 2012 data point is missing because Japanese statistics have now switched to 2010 weights. These are not comparable – 2010 weights generally imply lower inflation than 2005 weights.
@ Robert, Thorstein and others: Regarding the sectoral and open economy dimension. In Appendix J of the paper I extend the basic model so that imported oil is used as an input in production. This variant of the New Keynesian model predicts the same results as the baseline model. A reduction in foreign oil supply raises domestic real oil prices, which increases the marginal costs of production and in turn generates higher expected inflation. Since this lowers expected real interest rates at the ZLB, the negative supply shock is expansionary in the model.
@ Basho: Several papers have shown that consensus forecasts are typically better predictors of inflation than other (model-based or econometric) methods (e.g. Ang, Bekaert, and Wei, 2007).
@ 2slugbaits: In the paper I show that the supply shocks I analyze are temporary (so the second condition is satisfied). The third condition (“emergency conditions”) I analyze under the term “spillovers.” They key point of this provision is that the negative supply shock does not make the economy exit from the ZLB (i.e., the supply shock does not “spill-over” into normal times). I also show empirically that this provision applies.
jason11: Please note that the graph starts in 2010. The growth in Japan you refer to came before the earthquake, not after.
Posted by: JDH at June 27, 2013 03:17 PM
—————–
Does it? What do we call what happened from 2011Q2 – 2012Q1? Am I misreading that chart?
Looks growthy. Perhaps even rapid if you set your bar low enough.
It doesnt make any logical sense to think the inflation caused by a supply side event would be able to go back in time and erase the supply side event thats causing the inflation spike.
Pre-earthquake the consensus expected slow growth. Post earhquake they expected somewhat faster growth. The actual data shows more rapid growth than originally predicted. The level of production is lower, but the rate of growth is rapid.
JFW: Thank you for taking the time to respond to my rather off-the-cuff and flippant comment. Others are better equipped than I to directly engage the detailed merits of your paper. It’s not close to my core expertise.
I will say that I would not have intuitively expected oil shocks to be expansionary for oil importers even at the ZLB, and that I do not consider results based on oil shocks to say anything general about the effects of increased inflation at the ZLB.
I would like to expand a bit on my comments about Japan. I believe you noted that you had considered Japan’s earthquake/tsunami/meltdown event as a “temporary” shock, and that this assumption was pertinent to your use of the event. While there were certainly many short-term effects, I consider the most significant feature of that event to have been a permanent (or at least indefinitely extended) shift in Japan’s structural energy markets. Japan had, prior to the event, established a significant fraction of domestic energy supply from nuclear power, which allowed them to substitute labor and capital for scarce domestic resources (fuel is an insignificant cost factor in nuclear power) and was a huge factor in their trade balance up until the event. After the event, which damaged less than 10% of the nuclear power supply, all of the nuclear supply was shutdown, and it is still unclear whether these plants would ever be returned to service. This, and not damage from the tsunami, is the reason for an extended electricity shortage. The shortfall has been partially made up by continuing massive increased imports of fuel oil, crude oil for direct combustion, and LNG for combustion in the thankfully rather overbuilt Japanese electricity generation sector. Had the Japanese electric sector suffered massive shutdowns of primarily gas/oil fired plants, without the nuclear meltdown, and if the alternative hypothetical effect of loss of those plants had been no increase in fossil fuel imports and an increase in powerplant construction to replace, the effect on GDP would have been rather different (although unless the replacement plants were markedly superior to those they replaced, I’d have less to say about the overall net benefit).
@JFW. Still didn’t answer my contention that the Japanese earthquake should have and did affect the demand side as well.
Eggertson’s model is closed economy — which makes sense b/c there was little trade by 1933.
The result in your appendix is, I believe, a result (in part) of your CES preferences. The price of oil rises and total expenditures on oil is flat, thus it doesn’t crowd out other consumption. Higher oil prices also implies a larger trade deficit and thus capital account surplus and higher debt/GDP. You don’t discuss any of this, and I notice that your model has one agent and not two like Krugman/Eggertson. In any case, terms of trade shocks and supply shocks are not the same. Troubling that the introduction of the paper doesn’t seem alert to this.
I’d agree with you that Eggertson’s counterintuitive supply-side result doesn’t apply now for different, and more obvious reasons. But what you’ve done is snipe away at Keynes by disproving something nobody asserted with poorly chosen examples.
@ benamery21: […] and that I do not consider results based on oil shocks to say anything general about the effects of increased inflation at the ZLB.
I disagree. If inflation is indeed as helpful at the ZLB as some have suggested, then shouldn’t oil shocks, which are a non-trivial source of inflation, also have some desirable effects at the ZLB? In fact, if you look at the paper I show that oil shocks are more contractionary at the ZLB than in normal times, even though that is exactly the time where inflation is supposedly helpful.
I would like to expand a bit on my comments about Japan. […]
I discuss the permanent vs temporary issue in section 3.2 of the paper. The bottom line is that even if the earthquake had some permanent reallocation effects (e.g., electricity production – although that is not clear yet), the permanent effects are too small to explain the large short-run contraction.
@Thorstein. Still didn’t answer my contention that the Japanese earthquake should have and did affect the demand side as well.
I hadn’t seen this post when I sent off my post (moderation lag?). I absolutely agree with you that the earthquake had demand-side effects. In fact Sections 5 & 6 (p.22-35) of the paper model endogenous demand effects from supply shocks and explore their policy implications.
The result in your appendix is, I believe, a result (in part) of your CES preferences. The price of oil rises and total expenditures on oil is flat, thus it doesn’t crowd out other consumption.
In the model total expenditures on oil can go up depending on the elasticity between oil and labor (less than 1 means higher oil expenditure). The results are qualitatively the same either way. I can also write down a model where consumers need a fixed quantity of oil, i.e. need to spend more on oil after prices rise. Then the remaining consumption basket is subject to the Euler equation, and again that consumption will rise because real interest rates fall. In Euler equation models, crowding out generally comes from the real interest rate. If it falls it’s hard to make crowding out work (but see Section 5 of the paper).
Higher oil prices also implies a larger trade deficit and thus capital account surplus and higher debt/GDP.
The incomplete market extension incorporates this, although it is not discussed explicitly.
I notice that your model has one agent and not two like Krugman/Eggertson.
Sections 5 & 6 (p.22-35) show how a model can match the empirical facts. It has two agents — borrowers and lenders — as in EK. But the policy implications are different, because this model was designed to match the contractionary effects of supply shocks (which EK is not).
In any case, terms of trade shocks and supply shocks are not the same.
I use data on supply shocks and therefore model supply shocks. In standard (open and closed) NK models these supply shocks have expansionary effects at the ZLB. Of course, in open economies these supply shocks also affect the terms of trade. I haven’t claimed that terms of trade shocks and supply shocks are the same.
I’d agree with you that Eggertson’s counterintuitive supply-side result doesn’t apply now for different, and more obvious reasons.
This is one of the main points of the paper: If the counterintuitive supply-side doesn’t apply then we cannot expect large policy multipliers (fiscal/monetary) through the inflation expectations channel.
But what you’ve done is snipe away at Keynes by disproving something nobody asserted with poorly chosen examples.
Nobody? See the opening quote of the paper. I think the tone of the paper is very measured.
@JFW — I also remember Krugman himself posted about the impact of higher oil prices in a liquidity trap and concluded they would be contractionary, though perhaps not as contractionary as usual…
Another economist who hasn’t actually been reading Krugman over the past few years…
This to me is an effort to prove the utterly obvious, but apparently for some people it’s controversial.
The models that exaggerate the importance of real interest rates forget …
… consumers don’t follow the aggregate inflation and benchmark interest rates. Consumers react to their particular estimation of potential income from potential savings versus their particular expectation of inflation on what they would like to own. These expectations are likely to be vague, uninformed or out of date. Imagining that consumers would react to rising oil prices by buying more clothes flies in the face of common sense.
… while rising or falling real rates can be an impulse to delay or bring forward consumption, it depends very much on the reason. A price spike due to a shortage might prompt some people to hoard in case the shortage deepens, but that depends on the situation and on the practicalities of hoarding (eg hoarding salt is easy, food somewhat harder, gasoline downright dangerous).
… consumption is quite resilient to rising real interest rates. This has been evidenced most strongly by the consumer electronics boom since the 1980s, which has happened despite strongly deflationary prices (those who save to buy tomorrow’s consumer electronics rather than buying today’s enjoy very high real rates on their savings, but that has proven to be not much motive to delay consumption).
… the responsiveness of consumers to falling real rates depends on their access to credit and perceptions of the riskiness of debt. For example after 2007 access to credit plummeted and perceptions of riskiness of debt skyrocketed. Generally consumption of durable goods and investment in homes – credit-financed spending – are by far most responsive to falling real rates.
Those are just a few examples. I could go on for a while, but no time now.
@Thorsten I also remember Krugman himself posted about the impact of higher oil prices in a liquidity trap and concluded they would be contractionary, though perhaps not as contractionary as usual…
If you would actually read my paper you would see that in the data oil shocks are more contractionary at the ZLB, contrary to this claim. Maybe Krugman doesn’t get everything right after all? Also, if you look at yesterdays column he again claims that supply shocks (tighter environmental regulations) will not cost jobs in the current situation, because only demand matters. However, when supply shocks have demand effects — as they appear to do in the data — this neat separation no longer works. (Note: these policies may or may not pass cost/benefit analysis in any case — I don’t make any claims about that).
Another economist who hasn’t actually been reading Krugman over the past few years…
On the contrary, I actually have great sympathy for Krugman. Especially when he complains that the people arguing with him don’t actually read what he writes.
@Tom I also go through a long list at the beginning of Section 5 of the paper. I don’t think it is an effort to prove the obvious since some people have been making claims along these lines. Furthermore, once one acknowledges this “obvious” fact, then this has important implications for how effective we think demand-side policies are at the ZLB. In particular, in Section 6 of the paper I show that we shouldn’t pin our hopes on the inflation expectations channel.
Japanese nukes produced ~30% of the country’s electricity prior to the tsunami. That declined to zero. 96% of fully functional Japanese nukes remain off today. The direct structural increase in fossil fuel imports would directly account for ~25% of the GDP decrease you mention as projected for 2012. While I certainly hope they don’t remain off, it’s not looking very temporary.
Also, given the greatly reduced capacity of the grid, electricity rationing was instituted, rather than real-time-pricing. This has certain problems for your premise, in my view.
@JFW — Sharp comments, and nice defense of your paper.
@benamery21 — The author is arguing the Tsunami was a supply-side shock. It is only problematic if there was also a direct hit to demand — such as if Japanese began cancelling their vacations en masse, which they did, as evidenced by Hawaii’s excellent tourism data.
@JWF — the blog post represents the Japanese tsunami as a test case of what happens when productive capacity is destroyed, not as a shock to consumer spending. You seem to be saying now that either your mind changed or the blog post misrepresented your paper.
Again, the second example in your paper is about a terms of trade shock, which I see, and certainly Paul Krugman sees and Gauti Eggertson would see, as being quite different from a pure supply shock. But the abstract of your paper and the Krugman quote on page 1 (sorry, didn’t read past that) are about pure supply shocks, while the evidence you present is about terms of trade shocks. Though they may appear similar, these are actually quite different animals…
It’s certainly possibly to tweak your model slightly in the direction of plausibility and reverse your result. For one, in your model, even as the price of oil rises, spending on each individual consumption good as a share of consumption spending is fixed by assumption, owing to your unrealistic CES preferences. If a group of consumers are cash constrained, nothing that happens to interest rates in the Euler equation is going to overturn the fact that this group of consumers spending on other goods will fall. Thus the overall impact should be ambiguous. That you don’t/didn’t see this is not your fault — in grad school you got so used to writing down euler equations in which slight changes to interest rates impact consumption that you probably didn’t realize this was a problematic assumption. There are probably another dozen ways to move in the direction of plausibility and overturn your theoretical results while leaving the supply-side results in check.
Having said all this, I’d agree that, right now, even pure supply-side shocks are contractionary. It’s been clear since 2009 that the Fed isn’t powerless, and there are still plenty of interest rates in the economy that are well above zero — including interest on reserves. This Fed is guaranteed to overreact to any slight uptick in inflation…
Thorstein: I see state-imposed rationing of electricity as a demand shock–at least the rationing in non-industrial sectors. Also, I see it as creating allocation inefficiencies which are not necessary consequences of a different supply shock(not saying there were feasible alternatives in the specific case) which are causing GDP damage thru a channel other than inflation and which cannot be generalized to cases not involving rationing.
I see energy as fundamentally different from other goods, anyway, which I believe calls into question criticism of a general model on the basis of specifically energy related shocks.
Further, JFW assumes the Japan shock was not persistent, hence the details and magnitude of the continuing nuclear shutdown seem pertinent.
JFW In the paper I show that the supply shocks I analyze are temporary (so the second condition is satisfied).
Ummm….no. The Japanese earthquake shock is still ongoing. But I probably wasn’t clear. In the Eggertsson model the supply shocks must be BOTH temporary and have a well known end date. None of the supply shocks described in your paper had a known shelf-life date when the supply shock happened. In his New Deal example he quotes from the legislation that gives a definite end date. Now this probably means that, as a practical matter, the only kinds of supply shocks that would fit the Eggertsson model are conscious policy shocks. So what this says is that the Japanese earthquake example is probably not a good example of the Eggertsson thesis even though Krugman (and perhaps Eggertsson himself) tried to fit it into the thesis.
The third condition (“emergency conditions”) I analyze under the term “spillovers.”
But I believe Eggertsson’s understanding of “emergency conditions” was quite different.
The 2012 data point is missing because Japanese statistics have now switched to 2010 weights. These are not comparable – 2010 weights generally imply lower inflation than 2005 weights.
Agree. And that is my point. And this point is further reinforced by the implicit GDP deflator. Actual inflation has not matched forecast inflation from April 2011. And I’m pretty sure that the Japanese figured this out between April 2011 and the end of 2012. As Eggertsson argues, it is critical that actual inflation “accurately” match expected inflation. Clearly that did not happen, so we should not be surprised that GDP falls off in late 2012 as Japanese consumers begin to realize that inflation just ain’t happening.
A couple of points. First, Eggertsson’s argument is about “excessive deflation” (he uses that phrase several times) following “persistent deflationary shocks.” He also talks about “deflationary spirals.” None of these conditions applies to Japan. Japan has certainly experienced deflation, but nothing like the kind of deflation that informed Eggertsson’s analysis of the New Deal. Second, in the case of the Eggertsson model it probably makes more sense to look at the GDP deflator rather than the CPI and CPI expectations. And that leads to my third comment. Trying to fit this kind of analysis into an NK DSGE framework really isn’t convincing, regardless of which side of the argument one is presenting. It’s not consumer expectations of inflation that drive possible positive GDP shocks from negative supply shocks. In that regard Eggertsson’s transmission mechanism is unconvincing. I think a better theory is that with certain kinds of positive supply shocks in a deflationary world businesses realize that they have a window within which they can temporarily lock in cost inputs (e.g., existing contracts) and increase production in anticipation of higher prices one period ahead. In some ways similar to an investment tax credit. Businesses pull forward production in period t(0) know that in time period t(1) they will be able to extract some rents because of higher prices. And that’s why the GDP deflator is a better measure because it better captures domestic production costs. The CPI includes a lot of imports, which are clearly inapplicable to the Eggertsson model.
I am perfectly happy to discuss the merits of my argument, but I find this discussion quite frustrating. It is evident that you dind’t actually bother reading the paper, which addresses approximately 99% of the points you raise here. The reason there is a paper (and not just a blog post) is precisely to address these issues. If you think that you can form an accurate assessment of the full merits of the paper based on a blog post, then I am afraid you are wrong. Blogs highlight key findings. They are not a substitute for reading the paper! If you don’t find the blog post convincing, then you should read the paper. There is a reason it is 100 pages long (including appendix). It is not my job or desire to spoon-feed you the results. Life is short and time is too valuable.
I am very willing to respond to questions that deal with the substance of my argument, but I will only reciprocate the amount of effort that was put into any comment (0 = no reply). Or I may just post the entire paper in the comment thread 🙂
Congrats and Good luck in the new position at UCSD. I hope we’ll hear more from you here.
@JFW I considered that you may be correct, and gave your paper another look. It seems you have a very broad view of “supply shocks”. In your paper you very clearly lump together terms of trade shocks (oil), and random shocks to consumption (part of the effect of the tsunami) with pure “supply shocks”, such as from farmers colluding to raise prices.
But I don’t think you can say Japanese suddenly cancelled their vacations en masse (for those just joining us, see the link to Hawaiian tourism data post-tsunami above) only due to uncertainty over future income, and label it a pure supply shock… I’d go one further and say that when people suddenly stop spending due to uncertainty you shouldn’t wait until mid-paper to say that this is what your paper title had in mind by “negative supply shock”.
Thorstein Veblen: I agree that the aftermath of Fukushima had effects on both supply and demand. But please note that the event seems to have resulted in higher aggregate inflation, an outcome inconsistent with your insistence that the shock operated primarily through a decrease in aggregate demand.
@JDH Thanks for your response. I think we agree that a drop in demand wasn’t the primary impact of Fukushima, and also we’d agree that Fukushima turns out not to represent a test case of a pure supply shock.
I was interested about how inflation responded, so I looked it up: http://www.tradingeconomics.com/japan/core-inflation-rate
And, you are correct — after the tsunami, core deflation decreased sharply in Japan, especially in Q3, 2011, which was also the fastest recorded GDP growth Japan has experienced in two decades. I think that blurs the plot line a bit, once again…
@ Thorston: I don’t understand why you insist that global oil supply disruption are ToT shocks? Sure they affect the terms of trade in open economies, but that is an endogenous outcome and not a shock. In any case, what the NK model tells us is that foreign oil supply disruption and domestic supply shocks should both be expansionary at the ZLB through the inflation expectation channel. It that sense, I think it is perfectly permissible to use these shocks as a test of the theory.
Also, and I am very explicit about this in the paper , I think it is absolutely crucial to understand the demand effects of the earthquake (just read the intro to section 5). But to re-emphasize: these demand effects are an endogenous outcome of the supply shock, not an exogenous shock! But if supply shocks have demand effects, then we can no longer say supply doesn’t matter at the ZLB.
Finally, I think one needs to be clear about growth vs levels. The NK claim is that the level of output is higher after a negative supply shock at the ZLB — and growth is in fact slower as the economy converges back to balanced growth. The fact that we see lower levels and higher growth contradicts this claim on both counts.
@Anonymous
As I wrote above, domestic supply disruptions and oil price shocks are different. NK models don’t necessarily tell us they both are always expansionary. Namely, if domestic farmers collude to raise farm prices, this will raise farmers income, which means the nominal income of (at least these) domestic residents increases. This will be spent or saved domestically. In an oil price shock, the oil producers nominal income rises, but they are foreigners. The trade deficit should increase, which equals I – S, and since oil is also used in production, it’s unlikely Investment will increase. So savings falls. This is rough if you are in a balance sheet recession…
In an Nk model you have the same income incidence that you describe yet it doesn’t matter as I show in the appendix to my paper.
I’m curious to know how exactly your model works — especially what the savings problem is and which way exchange rates move. Would love to see it written down formally…