Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=Transcript of the talk, from the video's captions. Auto-generated: speaker names in particular are unreliable.
=

# What if Germany is cut off from Russian gas? (Markus' Academy, 7 April 2022)

Authors:   
Discussant: None  
Video: https://www.youtube.com/watch?v=6WQeAJk_390&t=0s


## Talk (00:00:00 – 01:30:47)

[00:00:04] welcome back everybody for another webinar organized by princeton for everyone worldwide we're very happy to have david bakari and ben moore with us hi david hi ben david and ben will present some work  
[00:00:20] they did together with many others from germany and titled what if germany is cut off from russian oil and gas and studying essentially implications on the german economy so we all talk about sanctions we're all  
[00:00:34] worried about the situation in ukraine and uh let's think more deeply you know how this will impact the german economy if there's a cut-off of russian energy first i've listed here the names of the  
[00:00:49] authors so to give credit to all of them and david and ben are here for them and what they point out correctly is that you know coil coal oil and natural gas are very different animals because  
[00:01:03] coal and oil can be created on the world market so what you do is you know is just substitute with other suppliers and they also has a different implications for russia russia can just sell to other  
[00:01:14] customers so if china and india are willing to buy the oil from russia then it's not big not so costly for them and it's also not too costly for the west western posted sanctions the situation is very different for gas  
[00:01:29] because it's more local it's more regional depends on pipelines so you have to find some different energy sources so you can't just substitute with different suppliers and you need terminals um and to get the liquefied natural gas  
[00:01:43] uh into the right place so it's a more difficult situation so situations depends very much what source of energy it is but essentially when you have a low cost of sanctions there's also low effectiveness because the other party  
[00:01:57] can get around it if they're high costs then there might be also a high effectiveness now the big question among the others is should we do it or should we not do it what are the implications but there's also dynamic dimension to it do you want  
[00:02:12] to have some so-called blitz sanctions so you just call cold turkey right away uh very harsh that will be most effective as russia cannot adjust  
[00:02:24] but it will be also most costly for the west as the west cannot adjust and essentially it's most effective because russia cannot redirect all the supply to  
[00:02:38] other countries and it might lead to a quick military withdrawal if one hits very quickly very hard that's one uh interpretation but it might also mean if it doesn't work then it means in the long run the  
[00:02:52] sanctions might be less sustainable and the question is do you want to build up some reserves to sustain sanctions in the long run and you might be worried you know if trump is re-elected in 2024 and is less interested in nato  
[00:03:06] uh you know wars typically can last decades and that's essentially uh you know something to think about as well so this there's some trade-offs here to consider and it's not obvious but of course it goes beyond  
[00:03:20] economics to make a judgment call on that now in general in economics we can think of two studies in the first one is a macroeconomic approach you really try to estimate substitutability across sectors  
[00:03:34] and try to how easily can you get free yourself from natural gas in particular and replace it with other energies and also replace the products which are very gas dependent with other products an alternative approach is a much more  
[00:03:49] detailed approach where you look at the physics directly uh you know look at the gas pipelines how it's transported from one spot to another spot can you bring it there or not it matters how the gas pressure is in the pipelines if the pressure drops too  
[00:04:04] much it might not work anymore so you have a much more physical detailed approach and next week we will do and going these details uh with who is actually an expert on sanctions  
[00:04:18] has studied sanctions uh for many many years and uh will go also into financial sanctions so we will have more than next week on this as well today we will focus on the macro approach but of course we have to keep in mind there  
[00:04:31] might be unintended consequences i was told that the ukrainian diesel for the ukrainian military comes mostly from poland while poland's to make diesel comes from russia so essentially if you cut off the russian  
[00:04:45] oil to poland it might be harder to supply the diesel for the ukrainian military so there are a lot of details which need to be figured out but today i think today's approach we will do a lot on subsidiability so let me just  
[00:04:59] bring everybody on board uh just highlighting few simple points uh before we give the floor to ben and to david david so here what i have i have on the x axis how much gas you have to produce something and you  
[00:05:13] can substitute it for some other input and they have drawn an iso equined which essentially says if you want to produce the same amount that's you know how much you can move things around how much you can substitute so if you move along this  
[00:05:27] white line you cut back on the gas input you have to increase the alternative substitutes by this amount so that's essentially given this production which is represented by this isoquant here you can see that the  
[00:05:40] reduction in gas in order to compensate this to get the same output you have to increase some other input by this amount now if you have a different production technology like represented by this  
[00:05:52] green isoquant so which is you know more curved but then if you cut back the gas input by the same amount you have to input much more of this substitute than compared to the blue one so the  
[00:06:06] green one has lower substitutability compared to the blue one and of course you can go very extreme and the very extreme version is when you have what's referred to as a leontif  
[00:06:19] production functions then you have actually a kink here so it's like if you want to cut back a gas you can't do it at all because you need an infinite amount of the substitute and this way you cannot really substitute it well  
[00:06:33] as a way at all now my first comment is that you know when we estimate this elasticity of substitution so how much substitutability do we have uh you know how curved are the isaac ones how much  
[00:06:48] do we need to replace them like what we had in this blue one we might be in a situation where the situation is curved slightly curved and we're around the point we are but at some point it becomes very steep  
[00:07:02] so the estimate the local estimate might be not a good guidance what it happens after a big shock or put it differently there are no linearities in the elasticities so that's depicted here in this and this green line is that one  
[00:07:17] here so it's very normal around the point where we are but if you move further away then it becomes like a leontif very steep and so then it becomes very different so it's very difficult to figure out you know when you make your local estimates  
[00:07:30] how they will hold for low for global large shocks and that might lead to some known resilience and sustainability we're saying oh actually there's some resilience there there's some subsidiability there but if it goes too  
[00:07:43] far it might not work anymore so to go back to some earlier webinar we had with jim hamilton one way to test for that is to use this method on the 1973 oil shock from opec  
[00:07:57] so just to give you some comparisons the current world supply for oil not for gas just simply for oil if coming from russia is 13 percent well and if you shut it down all the 13 are gone  
[00:08:12] if they're not substituted away you know move to china if they move to china then it's not it's a wash it doesn't change anything so rather than going to the west that goes to china and at the end it's not costly for russia it doesn't benefit anybody  
[00:08:27] perhaps china getting oil cheaper but essentially if you shut it off worldwide it would be a reduction by 13 percent the opec shock was an oil supply deduction by seven percent and that's the slide i took from jim hamilton's  
[00:08:40] presentation where he shows in 1973 how the oil production went down at the maximum drawdown was about seven percent so it would be interesting to see how things moved at that time how the estimates  
[00:08:55] would work at that time because we know how the economy reacted if we just do a counterfactual at that time and what would argue now in 2022 we're much less oil dependent i think we were in the early 70s but we  
[00:09:09] could also put the counter argument forward that we have already squeezed out the the last efficiency unit of oiling uh or any gas and other things so that to get that extra efficiency unit out it will be much more complicated  
[00:09:24] in this sentence now these elasticities are interesting the question is how do they aggregate and have a simple example just to show some complications which emerge so here if a production there is an upstream  
[00:09:38] firm and there's another upstream firm and then there's a downstream firm taking as input the outputs from the two upstream firms and then producing an output so in the simple example what i did is i just assumed there's leontive production  
[00:09:52] functions upstream in both uh upstream companies so that's there's no substitutability between gas and the other input x and there's no substitution between the input y and the gas in this this other sector  
[00:10:06] so you can't substitute at all but you know the firm and downstream can substitute between this firm and that firm's input so the output of this firm is an input for this downstream firm and  
[00:10:18] the output of this firm is an input for the downstream firm as well so here i assume the substability at the downstream firm now if there is leontive up there and if it's 50 50 let's say for  
[00:10:33] simplicity but then actually substability doesn't help much so you also get overall the aggregate elasticity of substitution is also leon tiff so it's also there's no substitutability on the other hand if you know there are  
[00:10:45] 60 40 here and there's 50 50 over there but then actually you can shift across the two inputs and this way you get uh substability in the aggregate so it's very subtle how the subsidiability aggregates over time the other thing i  
[00:11:00] would like to emphasize if you cannot move gas from this company and that location to gas to that location then it's actually a more problematic as well so one has to take this transportability of the gas into account  
[00:11:15] as well finally i would like to make a quick point about the production chains so we had already a very short production chain but we can also have a longer production chain so here is like a long  
[00:11:28] production chain and if you have an o-ring theory saying or if something goes wrong the whole chain goes down so if there's a shock at the top of the chain but then the whole thing might might be going down that's the question  
[00:11:41] is when we compare to the copic crisis the shock was at the final sector consume facing the consumer in energy it's much more upstream but it's not clear at all whether that's a drawback or an advantage so in a sense  
[00:11:56] it's a drawback because if you have an o-ring no substability up there and you can't substitute much at all it a little bit of a disruption kills the whole chain and makes the whole chain go away it has  
[00:12:09] much bigger amplification effects on the other hand if there is substitutability at each stage of this chain so you can also substitute at each stage of this chain as well and then it plays out very difficult so you can see  
[00:12:23] it's very subtle how the subsidiability aggregate over time and that david of course is an expert on this dimension now finally i would like to say something about financial afflictions and how financial frictions add to the  
[00:12:37] complications even further so you might have some adjustment frictions and ideally company a should actually scale back because it uses so much gas and company b should scale up because it uses some renewable energies and should  
[00:12:50] actually scale up its production but because of financial frictions company b can't get financing so it's not so easy to get the finance doesn't have the collateral or any other while company a because of this high gas prices will  
[00:13:04] actually go bankrupt so in an ideal world without financial frictions company a would just scale back and company b would scale up but what we'll have with financial fractions potentially this company a will go bankrupt scales back of course  
[00:13:19] dramatically and company b cannot scale up because it has financial fiction so an aggregate it might be much worse now let me go to the poll questions ben and david have put forward and these are the questions they were  
[00:13:33] asking you and the first question was following an import stop of russian energy so everything coal oil and gas everything together how much will the german gdp decline relative to not doing anything and just  
[00:13:48] leaving the economy running as it is uh now or was before any sanctions were imposed and the answers were um so less than one percent is seven percent let me go to the other extreme  
[00:14:01] more than 10 percent thought 10 people thought more than 10 percent uh the economy would tank we forgot to put the 5 to 10 percent in so i'm not saying anything but 1 to 3  
[00:14:13] was 36 percent three to five percent was 47 percent so you can see most people think it's in the range between one and five or perhaps above five and ten percent thought it's about ten  
[00:14:27] percent in terms of gdp now if we do alternatively we impose some 40 percent tariff on rational energy and had the same question again just instead of total ban uh called turkey let's do a tariff like what carter  
[00:14:41] hausmann proposed and we covered this earlier in the jim hamilton webinar then the answers were different and the answers going to the same numbers were 27 percent  
[00:14:54] said less than one percent so it's way more moderate forty one percent thought one two three percent twenty percent thought three to five percent and eleven percent thought five to ten percent and  
[00:15:08] um more than ten percent essentially only one person asked under that so it was very uh very unlikely that we would tank the economy more than 10 finally the final question was about  
[00:15:22] implications of doing a cold turkey off of all energy supplies from russia um about the implications on inflation less than two percent that's what 26 percent thought two to four percent  
[00:15:36] that's uh 53 percent of big majority about two to four percent extra percentage points on top of the inflation we have and more than six percent uh that's what's twenty percent uh thought this way  
[00:15:49] so um i live with that and i pass on the floor to ben and to david and they will present their results to us and we will have questions during the presentation thanks again to david and ben and we're  
[00:16:03] looking forward to the presentation great um thank you so much uh marcus for having us thanks everyone so much for coming uh really exciting to be here um so i'll do the first part and david will  
[00:16:17] do the middle and i'll do the conclusion so we'll switch back and forth a bit so as marcus already said this is uh uh joint work with a large team of uh sort of cutting across different subfields in economics that's why it's so many people in a sense um  
[00:16:31] okay so you know what are the objectives of this paper um i mean marcus has already said a lot of this but essentially our goal was just to assess the economic consequences for germany of a cut-off from  
[00:16:44] russian energy imports and there's sort of two reasons why you think that may be important the first one is uh either germany or the eu may implement an embargo you know from our side or i guess the  
[00:16:58] other reason why it's important is you know it's conceivable that putin will sort of switch off the gas tap from his side um our goal here was sort of to provide a  
[00:17:10] worst case assessment of sort of extreme assessment of a cold turkey complete import stop and uh you know i'll come back to this it's not because we think for example a full embargo is necessarily the best thing to do it sort of the idea was to  
[00:17:25] maybe bound other scenarios for example you'd think that maybe with a tariff that would be bounded above by the sort of cold turkey import stop and uh the other reason is that you know less  
[00:17:38] extreme policies like a tariff may then still trigger a full stop uh from the russian side okay the basic you know philosophy here was we wanted to get a sense of the rough magnitudes of economic losses um uh from  
[00:17:52] doing such from having such an input stop relative to a do nothing baseline and i guess here's some possibilities this is sort of similar bits to the to the poll that we had so you may think maybe there's small  
[00:18:05] small gdp decline uh say 0.5 to 1 uh relative to baseline given there's you know growth in the baseline um you could imagine maybe not even a recession uh maybe you think it's going to be like the covet recession that size of  
[00:18:20] magnitude in germany it was a four and a half percentage point a gdp decline maybe you think it's like spain or portugal during the euro crisis so we i looked this up it's like five and seven percent gdp loss and then the other scenario that could  
[00:18:34] potentially be the case um could be something like mass unemployment and poverty which for example in germany some politicians have used these words um on tv to you know warn about an import stop um which to me at least sounds uh something like a the great  
[00:18:48] depression and definitely a gdp loss of more than 10 okay so those are kind of you know we just wanted to have a sense of where are we going to be in there okay here are the possibilities again our headline numbers um in the paper are essentially  
[00:19:02] a gdp decline between half a percentage point in the most optimistic scenario and three percentage points okay so the takeaways here that we want to make is essentially an import stop uh will be  
[00:19:15] likely somewhat less severe than the covet recession which uh you know would be four and a half percentage points um and then uh you know we don't want to say you know we we should be doing this or we should not be doing this um but we  
[00:19:29] want to make the point that in principle we think germany is a rich country with a moderate debt to gdp ratio of something like 71 percent um you know there's fiscal capacity so uh you can potentially um you know do  
[00:19:43] something there and uh provide insurance and socialize the cost we did the same thing uh during colbit okay um here's something that's we're not going to cover in the paper that we wrote um but i'll talk about a little  
[00:19:58] bit at the end um which is uh uh the effects of an import stop on inflation okay just to you know manage expectations a bit um okay just to come back to this i guess to relate to the poll so i guess relative to the poll  
[00:20:13] we're uh sort of a little bit lower than the average uh poll respondent so i guess it's good that i'm presenting this we have some work to do to convince everyone that uh essentially uh you know we're sort of in the  
[00:20:25] more in the lower part of that uh uh range of estimates that people gave and definitely the people who said more than 10 percentage points we want to argue uh uh it's very hard to make that case okay  
[00:20:39] so um here's just a graph to show you um marcus already i think sort of had it uh german primary energy usage uh this is table one in the paper um these are the  
[00:20:50] main sort of uh energy sources up there in particular fossil fuels so oil gas and coal and then these other energy sources this is in terror with hours and then essentially the the main thing to look  
[00:21:03] at is how much of that energy comes from russia okay and you can see um especially for oil gas and coal these are large numbers in particular for gas here you have a whopping 55 of all gas in germany comes  
[00:21:17] from russia so you know you can see why people are uh scared to a certain extent okay as marcus has already said sort of um our argument is that to a certain extent um gas is really the the bottleneck the  
[00:21:32] tricky one and oil and coal or maybe not so problematic the argument is that oil and coal have a global market and germany has strategic reserves also oil and coal you can you know put on a  
[00:21:45] ship and move around much easier gas in contrast is really very much tied to the existing pipeline network um liquefied natural gas supplies in the world as a whole are just kind of small and you you cannot really substitute this too much as a  
[00:22:00] result uh we're going to focus here really for the paper uh almost exclusive exclusively on gas again because in germany at least that's sort of the big elephant in the room uh as you know how what would you do about gas  
[00:22:14] so ben yes can i ask you there's a question by uh he would like to know whether your analysis also applies to other european countries like austria and also to your east european countries in particular or central european countries as well and other spillover  
[00:22:29] effects from a german economy to let's say poland great question um we have a slide on this at the very end um we in fact did some analysis exactly for the eu as a whole we wrote a paper um for the french  
[00:22:44] council of economic advisors with uh camille londe and philippe natan where we did some sort of exercise exactly for the eu as a whole so let me you know bank the question um and revisit later but so your brief answer yes in  
[00:22:57] principle uh uh yes some such analysis also applies to the eu as a whole and uh these other countries okay so um so i said 55 of all gas uh comes from  
[00:23:12] russia um you know what's the size of the gas shock we want to think about so we want to say okay you do lose 55 of this gas but you know some of this you can substitute in in some way or another  
[00:23:25] okay and here we're relying on our uh friends the energy economists on the team and there's some very nice work by the people at uh brugel uh on this yeah um it's important marcus said already a little bit uh to think about time  
[00:23:39] horizons so i should have maybe already even said this earlier so for for most of our exercises we're going to want to think of the relevant time horizon essentially roughly until the next winter um why is that well that's because  
[00:23:53] there's a lot of seasonality in gas demand so in particular gas demand is much much higher in the winter when you know households in particular and also firms use it to heat a lot um so we already know basically um that  
[00:24:07] we're probably going to be fine over the summer um so the the question will be some you know how much extra gas will be able to get roughly until next winter and also um you know then going forward when we think about substitutability  
[00:24:22] along the production chain how much can you do there in terms of that type of substitution uh until next winter so we're talking time horizons you know nine months seven eight months uh uh definitely less than a year  
[00:24:35] uh the idea is you know some of these 55 percent here you can replace with imports from other countries in particular say norway or the netherlands some of these 55 percent are also used  
[00:24:48] in electricity generation um and uh from you know what i've been told you can essentially substitute that gas that's used in electricity generation with other energy sources so you can burn other  
[00:25:02] things so in particular ligna i'd say when our energy economist team members counted up these numbers and relied on some estimates in literature the number they came up with is essentially that you while you lose  
[00:25:17] 55 percent of the gas you can import a substitute 25 percent so that in the end the gas shock is essentially 30 okay so uh for all the following analysis essentially what we're going to work  
[00:25:30] with is uh the following energy shock here which is a 30 decline in gas or equivalently if you you know calculate that up to the total terawatt hours uh generated by gas oil and coal it's going to be sort of an eight or  
[00:25:45] maybe ten percent uh drop in in total energy or fossil fuels i guess just to clarify you assume that the three power plants which are still active until the end of the year will be extended or in in your analysis or not  
[00:25:59] i honestly forget the answer to this uh i think so but i'm not 100 sure um this is a question for and then the sanctions you if you look at the sanctions if they have to go over the next winter then the numbers will be  
[00:26:14] much larger so it's only if the sanctions last during the summer is this correct in your analysis no no no no no no no the the scenario we're always going to do is um you put the sanctions now and you put  
[00:26:27] them in place forever forever okay and then well or for forever and then let's say for a long time yeah yeah yeah and then it's a sort of uh the numbers i'm gonna say like are gdp loss say in the first year  
[00:26:41] um and then i guess in the second year you would get another uh gdp loss right but that in the second year you'd think that maybe that gdp loss would actually be less than in the first year why because you know there's going to be more you know adaptability and substitution as time progresses  
[00:26:56] you said roughly until next winter so so but you go over the winter then or not in your numbers sanctions are going to go over the winter yeah yeah no this is only sort of uh to say how much uh uh time uh  
[00:27:10] do we you know give ourselves here to you know make up for some of this 55 gas gap the point here is that the seasonality works in our favor right if now we're in the middle of winter  
[00:27:24] it'd be much harder um why because gas demand would be high so that's kind of the point here but yeah the experiment is always put the sanctions in place now and keep them for a long time it's not you know putting them in place  
[00:27:36] temporarily okay thanks for clarifying okay so here's the plan for the remainder of the talk um i wanted to give you some facts about german economy and its energy dependence um then sort of essentially starting from these facts we're gonna map this  
[00:27:50] sort of energy shock so minus thirty percent gas or eight or ten percent total energy into some aggregate uh macroeconomic losses so uh gdp or gross national internal expenditure and then we're going to use some macro models um there's going to be some different  
[00:28:05] models we're going to use there's going to be a very simple model essentially just the production function and this is going to serve to highlight the importance of substitutability mark has already talked about this we're going to spend in this  
[00:28:18] presentation quite a bit of time on something that we didn't spend too much time in the paper but we think you know we should maybe spend more time when we revise the paper which is david came up with this very nice sort of sufficient statistics formula for that  
[00:28:33] you can use in richer models in particular with supply chains um and international trade maybe uh and that's based on this work with uh emanuel fary and then we're going to show you some model simulations together with some back of the envelope calculations using  
[00:28:48] these statistics the sufficient statistics for these kind of models where you have rich supply chains and international trades then there's going to be some more general stuff for example um you know  
[00:29:01] just to discuss discussion of mechanisms outside of the model and other studies and i'm going to come back to the question that was already asked you know what about other countries and then i'm going to discuss also a little bit this embargo versus tariff question  
[00:29:16] okay so here's some facts about the german economy the first thing is or the first three bullet points here are just some numbers of you know how much of the total  
[00:29:30] economic pie in germany um how much of these sort of payments if you want go uh to these energy sources uh and the point is you know the the numbers aren't massive so you know total consumption of gas oil and coal is  
[00:29:44] something like four percent of gross natural expenditure um if you look at the imports of that that's less um why because for example some coal is produced domestically as well and if you look just at gas  
[00:29:58] where all of it is imported then it's essentially one percent of gross national expenditure goes to gas okay so these are not huge numbers these numbers might not be fluctuating a lot depending on their  
[00:30:11] like energy prices did you use what numbers did you use from doing the cobia yeah we use the that's that's a good question so we used the last number for a year for which there was data which was uh  
[00:30:25] i want to say 2020 or i think it was pre-coveted um i'm not sure um i should check this it's a good point we can take some historical averages maybe uh you know rather than relying on one year thanks  
[00:30:39] um okay but you know it's not gonna suddenly be five percent i think um okay here's some statistics on uh gas usage and across different sectors of the economy  
[00:30:52] sort of households industries so that's manufacturing and construction here services and electricity generation and so on and then i'm going to contrast this in a second with you know their economic importance in some sense you can see gas is kind of equally  
[00:31:07] distributed essentially across households um industrial use and then everything else okay so the one third one third one third i mean a lot of it is in electricity generation okay it's interesting to contrast this just  
[00:31:22] with how important you know these sectors are in in employment and production and you get uh numbers like this here um uh and so one reason why i wanted to  
[00:31:35] show this is if you if you just look at sort of the total employment that's in industry in germany um you know that's 22 so that's obviously large um but you know it's not 50  
[00:31:49] and in fact even in germany which is sort of people have in mind as a very sort of industrial manufacturing intensive country way more people work in services and so on similarly for gross value added you know the number is a little bit higher  
[00:32:04] but it's still sort of a quarter of the economy the numbers here obviously in bullet points one two three as i said are small um but you know you don't want to just conclude  
[00:32:19] from that immediately that the the aggregate gdp losses are going to be small say and i'll come back to this point why because obviously energy is a sort of a critical input in production and so you there's lots of amplification  
[00:32:31] mechanisms that could be important okay um i wanted to show you one more uh table with data and then go on to the modeling um this is some  
[00:32:45] statistics of gas uses and some other statistics across specific industries okay in particular focus on the left-hand side panel here for a moment and ignore  
[00:32:58] what's on the right-hand side here um this is uh you know the three sectors of the economy that we think would be hardest hit in the case of an import stop stock these three sectors  
[00:33:13] jointly make up for 59 of all industrial gas use which again was something like a third of total gas use these sectors are the chemicals industry um that comes up a lot in the discussion  
[00:33:27] it's the food processing industry and it's the metal industry okay um here's some other uh numbers here how many people work there for example and you can see the total employment here is maybe something like  
[00:33:41] uh you know three three and a half percent of gdp of total employment um and here's some other statistics um you know gross value added and gross output in these industries  
[00:33:56] we thought it was interesting to contrast the economic activity and employment in these industries which we again think would be hardest hit in an import stop with  
[00:34:09] the three industries that were hardest hit during the covet 19 pandemic okay these are air transportation and in particular hospitality and entertainment  
[00:34:23] and the point we want to make here is that if you look in terms of gross value added wages and employees roughly in total magnitude these sectors are kind of the same okay um  
[00:34:38] employees in fact um you know you can count up the numbers here um are actually quite a lot smaller in these three sectors than for example in hospitality is just big in terms of number of employees but if you look at  
[00:34:53] gross value added and things like how much capital there is they're kind of comparable [Music] there's a big difference in gender it's kind of interesting way more men working in uh these sectors here than uh  
[00:35:07] uh uh in these sectors um there's two things that are um obviously very important between the current crisis and or potential crisis in the case of an import stop and the 2020 crisis the first thing is  
[00:35:22] in the 2020 crisis we shut down these sectors completely essentially okay whereas now we're not talking about that right we're instead talking about losing 30 percent of the gas supplies here which then maybe in an extreme case  
[00:35:34] would shut down 30 percent of the production in these sectors you you'd think okay so we're not talking about completely shutting down these sectors um on the other hand what's very different is these are very sort of up sort of downstream sectors that are very  
[00:35:48] close to final consumption whereas these here are uh uh very upstream sectors so you'd think that if even if the shock is kind of small this can sort of propagate through the production network through the supply chain so banking are just compared with another crisis which like  
[00:36:03] the subgram crisis that was always argued you know the subprime losses but tiny yeah the banking was so central it dragged the whole world economy down too so it could be way more dramatic yeah yeah no no no i mean that's exactly the point  
[00:36:18] i was making here and i'll make a number of slides again but um yeah just you know just to put things into perspective a little bit uh uh you know i do think this uh there's something about this comparison that's useful  
[00:36:31] um okay and another worry that comes up a lot is uh uh distributional effects of a potential import stop um you know and the idea is that poor households  
[00:36:45] maybe have a much larger expenditure share um on energy and then if either the the you know gas get gets cut or the prices go up in particular these poor households would  
[00:36:59] be much harder hit there's a very uh uh nice survey data set in germany that actually has these very detailed questions about a household consumption of these different energy expenditures  
[00:37:14] and you don't find a very strong pattern across the income distribution there's a little bit of a gradient here you can see right um but overall you know this gradient here or this how the  
[00:37:28] expenditure share varies across the income distribution is relatively flat okay there's some other things that are interesting um you know it differs a lot more with household size rather than  
[00:37:41] you know just income of the households the two are obviously correlated um if you include car fuels which are not in this graph here then the graph looks a little different in particular it's actually sort of an  
[00:37:54] inverse u shape so a hum shaped with income why just because the poorest households don't tend to have cars as much essentially okay so we're we're not going to talk about  
[00:38:07] distributional effects uh a lot more um maybe we can come back to it in a discussion essentially we're going to say that um you know these are maybe something to worry about but it's not completely obvious if you look at the  
[00:38:21] data that these are going to be massive and you can probably uh compensate with uh with fiscal transfers if you're worried about them okay all right so let me uh come to the macro modeling part of things okay so again  
[00:38:35] what's the spirit here the spirit as always i guess in quantitative macro is we're gonna you know start from the facts the type of facts that i've just shown you um and then we're going to try to map uh this sort of energy shock that i  
[00:38:48] talked about uh this 30 drop or or if you think about in gas or if you think about total energy maybe 10 drop into um you know gdp uh or growth national income losses um using these  
[00:39:03] macro models and the the philosophy again right is going to be to try to think hard about specific mechanisms um like say these production chains or aggregate demand amplification and what have you and then try to put empirical  
[00:39:18] discipline on these as much as possible and then sort of try to see what comes out of it of course you know as mark marcus said um there's always a lot of uncertainty um about exactly what the  
[00:39:31] right parameter values are in particular a lot of these elasticities you have sort of very local estimates and then if you for a very large shock like this um you know there's there's some uncertainty how much you can sort of use  
[00:39:44] these these elasticies to extract late for these very large shocks but you know at in the end of the day it's still uh the best uh game in town i would say and so so that's what we're going to do here okay remember again just to remind you  
[00:39:59] that you know gas is something like one percent of g and e and the gas stock is 30 okay so i set the gas expenditure shares small but of course the issue right is that essentially substitutability may be  
[00:40:14] lower it may be a bottleneck in production okay just to you know have some sort of a a benchmark or comparison i want to put on the table two  
[00:40:28] calculations that i think are sort of the two extreme calculations you could do and they're also going to be i want to argue nonsensical calculations okay that you don't want to do because they're sort of just fly in the face of what we know about empirical evidence  
[00:40:42] okay so the first thing you can do um which again you don't want to do is you could say oh look like gas is one percent of gdp um it declines by 30 so we're gonna get a gdp loss of one  
[00:40:56] percent times thirty percent which would be zero point three percent okay um that's a crazy thing to do um why exactly because you know it would to completely ignore any of these sort  
[00:41:09] of o-ring or bottleneck or production chain effects and any amplification okay there's a very nice um article or interview i think by larry summers where he talks about the financial crisis this comes back to marcus's point just now um  
[00:41:23] and he likens the financial crisis to what if uh electricity were to go off okay and uh then he says okay electricity is only three percent or four percent of gdp um let's say electricity falls by eighty  
[00:41:37] percent um then he says there would be some crazy economist who would say you'd only lose uh three percent of gp i think he says uh although chicago and minnesota people would do this essentially um and  
[00:41:51] and he says that's a crazy thing to do and i i obviously we obviously agree with that okay um so you would think that it has to be larger than this on the flip side um uh the other crazy scenario to do would be  
[00:42:05] no substitutability whatsoever okay uh so gas is a complete bottleneck even though it's just one percentage point of gdp in that case what you would get right is that gdp falls essentially one for one  
[00:42:18] with gas and so if you have a 30 gas drop you would have a 30 gdp drop i want to say that's also a crazy view of the world essentially okay um now let's let's actually um you  
[00:42:32] know connect this with a little bit of economic theory um uh where you can sort of see this point nicely i think um and this is also going to be our sort of simplest model here which is simply just a ces production function okay  
[00:42:45] um so you have some output y that's produced using uh in particular gas and other stuff um say here's some aggregate of capital and labor but doesn't really matter some other factors for production  
[00:42:59] and the key parameters here will be two parameters so one will be the elasticity of substitution between a gas and other factors of production and the second one so that's the sigma  
[00:43:14] here and the second one will be the share of gas and production which is the alpha okay now uh it's relatively easy to see that the two i want to say crazy cases that i talked about here  
[00:43:27] above are exactly the two extremes of cop douglas production essentially in which case this calculation would be correct okay and uh leon jf so no substitutability  
[00:43:42] whatsoever uh in which case uh this calculation here would be correct so you know just to be clear in principle it is possible um to have uh that a 30  
[00:43:54] gas drop leads to 30 drop in gdp um if you really think that all uh substitutability is zero okay so ben you i guess you will open up this aggregate production function later on because yes the other thing is i want to say if  
[00:44:09] you have a 30 drop of gdp there would be political unrest on top of it that would be less even more than 30 percent i guess yes um yeah yeah and we'll yeah we'll do something much uh  
[00:44:22] you know richer than this here in a second okay um but i think it's so useful to give you to bound some magnitudes essentially okay um just some general thoughts but marcus  
[00:44:36] already kind of did a lot of this so i i'll go fast on uh elasticity of substitution and substitution more generally okay so one thing i think marcus already literally said most of these things first they're time dependent so you know  
[00:44:50] that in the short run elasticity or less than in the long run again i've already said what we want to think about here is a horizon a time horizon for substitution that's  
[00:45:03] roughly until the next winter why because we think sort of that's uh because of the seasonality and gas demand that's how much time we have so what are we going to do is we're essentially going to go to the empirical literature um where they estimate these  
[00:45:17] kind of elasticities um they typically uh distinguish between short run which is like less than a year and long run which is more than a year or much longer than a year we're going to take always the short run elasticities then we're always going to  
[00:45:30] go to the very bottom of the range of the empirical estimates and then we're going to kind of divide them by two okay to get basically the the sort of short-run analysis that we think are relevant and we're going to have very low elasticities but not zero  
[00:45:45] okay another thing that marcus already said is micro versus macro elasticities um so uh macro in particular takes into account not just substitution uh within a production process but also cross-production  
[00:46:00] processes or firms so there's maybe sort of an extensive margin so marcus had this nice example with the two leontiev uh uh technologies where you may still be able to substitute in the end  
[00:46:14] supply chains marcus i think had literally exactly these points so long supply chains create bottlenecks that's the o-ring point but on the other hand the longer the chain the more substitution possibilities so these  
[00:46:29] things kind of go both ways the final thing that's going to be important not in the cs example but in the richer models in a second is potentially substitution via imports so the idea is going to be okay so gas  
[00:46:44] is going to you know decline a lot and therefore maybe it's going to become a lot more expensive as well it's not going to be economically profitable to produce some intermediate goods anymore  
[00:46:57] but you could potentially replace these with imports okay so here's an example that comes up a lot in germany so there's a production chain which is something like this so gas is used to produce ammonia ammonia is used to  
[00:47:10] produce fertilizer and then the idea is if the gas declines we don't have fertilizer anymore um but you know what we would say is to a certain extent you can import the fertilizer instead and then still preserve some of the downstream  
[00:47:24] production of course it's then bad for the german fertilizer producers and and the jobs that are in that sector however it doesn't mean that the entire system that comes after that collapses but does your analysis take into account  
[00:47:38] that the fertilizer world price worldwide might be much more expensive yes we do take that into account yeah yeah yeah i'll come back to that in a second um or david will in fact we wrote a little supplement here on  
[00:47:52] substitution that makes some of these points and also has some uh historical examples of how economies respond to large shocks there's some interesting stories in there for example about airplane production in world war ii  
[00:48:05] where in the us um uh you know president roosevelt said we need 50 000 planes right now everyone uh the economist the industry people everyone said it was going to be impossible two years later they produced  
[00:48:17] a hundred thousand airplanes um here are just in the simple model so the cs production function output losses for different analysis of substitution okay  
[00:48:30] here i did a 10 energy drop which is if you think about oil gas and coal combined that's sort of the right magnitude so you start with energy  
[00:48:43] equals hundred percent so one here um and so you say i normalized production here to one if you're in the leontev case where the associative substitution here is  
[00:48:55] zero so the blue line here then as i've said you know if you have a ten percent energy drop you're going to have output dropping by 10 percentage points okay so you go 90 energy means 90 output okay  
[00:49:10] if you go cub douglas instead you get this much smaller number which is essentially just the energy share times the shock and then the question is sort of where are you in between okay what's  
[00:49:24] interesting i think and i didn't appreciate enough before you know making these calculations is that um because this gas share is small okay even with very low elasticities of  
[00:49:37] substitutions the output losses are potentially actually quite far from leontif so this is for example the red line here is an elasticity of substitution of 0.04 so quite a low number um  
[00:49:51] and you can see that the output loss here is still less than two percentage points essentially whereas with full leontif it would be ten percentage points okay we'll come back to this so it really matters you know what you think  
[00:50:04] about the elasticities uh how how small they are essentially okay uh and now i think uh david is going to take over uh great uh thanks a lot marcus and ben  
[00:50:17] um so um i think the way that we want to think about these economies though is that they're much more complicated than the simple um single ces nest would suggest so this is  
[00:50:30] what marcus was getting at real life economies are very complicated machines there's domestic supply chains that run across sectors and there's international supply chains that link these sectors to other producers in the rest of the world so  
[00:50:44] this picture that i have here is sort of an illustration of what a modern industrialized economy looks like each little dot in this picture represents an industry so you could take one of these dots is for example the  
[00:50:57] chemicals industry it's connected to other industries in the domestic economy because it's both buying inputs from the other industries and selling inputs to the other industries and it's connected to the rest of the world through imports  
[00:51:10] and final outputs where we think of exports as being part of the final outputs that these producers are producing so as this come up both in marcus's discussion and in dense discussion when you look at these complicated structures it's not  
[00:51:25] immediately obvious which way it's going to go relative to the simple model on the one hand the fact that you've got vertical chains pushes in the direction of maximizing the damages when there's a shock but on the other hand to the  
[00:51:39] extent that you have horizontal connections that opens up the door for substitutability where you can escape the shock further downstream um now we want to ultimately be able to think about economies that are complicated  
[00:51:54] like this um uh so ben if you could go to the next slide um and my starting point uh whenever i'm trying to think about these complicated economies is just to start off by thinking about maybe some sufficient  
[00:52:08] statistics intuitions that are going to apply across the board to a vast class of models just so i know what are the things that i need to be thinking about in order to understand the answer to the question that we're interested in so in this context we're interested in  
[00:52:22] two key aggregate statistics one is going to be a measure of german real consumption of resources so that is going to be real gne gross national expenditures and the other statistic that we're going  
[00:52:36] to be interested in is real production by german producers which is real gdp now these two things oftentimes get conflated in sort of the non-academic non-technical discussion but they're not actually the same thing the way to think  
[00:52:50] about it is that gdp includes things like the production of exports so when germany makes a car and ships it overseas that's going to be inside gdp and on the other hand gne includes consumption of imports so when german  
[00:53:03] consumers buy something from foreigners that's going to be inside gne but it won't be inside gdp and so these objects because of the way that they're exposed to international trade in different ways don't have to behave the same way  
[00:53:17] now our starting point for thinking about this what happens in a complicated economy like the one i showed you before is we're going to assume to start with that the economy has productive efficiency so the german production  
[00:53:30] network is efficient um and then if you can go to the next slide then then using that efficiency we can actually come up with some very general formulas for how german real consumption  
[00:53:45] and real production are going to be affected by a shock like the one we're talking about where you cut off imports from russia so what i'm showing you here is the beginnings of a approximation to the change in  
[00:53:58] real consumption by germany in response to basically any kind of shock you put in uh into your model so on the left-hand side we've got the change log change or if you like the percentage change in german real consumption and on the right hand side  
[00:54:13] i'm going to show you the different pieces of how real consumption is going to be affected by shocks so the first chart we've got here is just what ben was talking about which says if you reduce import so here m is imports  
[00:54:26] j is indexing the nature of that import so this could be for example russian gas or it could be um something else that you're getting from russia you take the shock to um the quantity of the import so in our example this is  
[00:54:41] going to be like a 30 reduction in your ability to source natural gas and you multiply it by the amount that germany collectively is spending on gas as a share of total consumption or total national consumption national  
[00:54:55] expenditures so that's sort of the first order term that uh ben was talking about and then if we can go a couple more slides so the next two terms here on the top line the first one is the  
[00:55:08] change in german exports and it's weighted by the expenditure share in g e just as german imports would be and the final term is the change in employment weighted by the  
[00:55:22] wage bill of the workers who are becoming unemployed so this is a first order approximation that works as long as the initial german production network was efficient and so one way you can think about it is if you  
[00:55:35] think there's strong keynesian effects or aggregate demand externalities or financial frictions that are going to result in reductions in employment that last term is what's going to capture that effect to a first order you have to think about how many workers are losing  
[00:55:48] their jobs and then what was the wage bill going to those workers in the initial equilibrium before the shock now if you can get one more perfect so the term that we've got now on the second line are the non-linear effects of these  
[00:56:03] shocks and these are really the things that ben was talking about and these are the things that complementarities and substitutabilities are going to play an important role in disciplining so what the terms in the square brackets  
[00:56:16] are is they're basically taking into account the fact that when um the supply of natural gas goes down in germany you would expect expenditures on natural gas in germany to go up as a share of  
[00:56:30] total expenditures and the fact that the expenditures are rising is going to mean that the reduction in the quantity of natural gas that you have access to becomes more important and becomes more costly  
[00:56:42] so these deltas that you see for example the first delta on the imports term is telling you you have to think about what is the change in expenditures on imports in response to the shock and if you can  
[00:56:56] come up with some either historical estimates of what you think that those changes are or uh you have a structural model with detailed um assumptions about how production works you can put a number on that but as a starting point i think it's very useful  
[00:57:10] to just understand that whatever number you're going to come up with at the end of the day it has to map into a sensible change in the expenditure share on imports for german now the leon tfk is just using this equation you can already think about  
[00:57:24] what's going on in the cob douglas case and at the leon tf case in the simple example that ben was showing you in the cobb douglas case the change in those expenditure shares is zero and so everything in the square brackets is  
[00:57:38] zero you don't have to think about that at all in the vyontf case what actually happens in response to a small shock to energy is that the expenditure share on energy the model would predict should jump to  
[00:57:52] one because the marginal value if you need energy to produce stuff it's like everything else is worthless now and so the only thing that you're going to want to spend money on is energy and so the model the leon tf model has these very  
[00:58:06] extreme predictions about how expenditure shares should move and how relative prices should move and i think that's why we think that's not realistic is because if you look at different historical episodes where there's been energy crises either in the 70s or in  
[00:58:19] japan where there was a fukushima you don't see such extreme changes in relative expenditure shares and so your beliefs about the relative expenditure share are going to discipline and how big do you think the effects are going to go on more  
[00:58:33] um and there's an equivalent um sort of type of result for real gdp but i'm going to focus we've focused on real consumption uh because that's a more relevant welfare measure uh in our notes uh next slide please great so what are  
[00:58:47] the key uncertainties what are the key quantities that we need to be able to put numbers on in order to get um a final answer i'm just going to go through the ones that we think are really important that you need to take a stance on um so you have you know different import  
[00:59:06] sectors different export sectors and so forth if i make it finer and finer does the approximation change i suppose i put them together or make them finer and finer yeah thousands of inputs and  
[00:59:19] thousands of expert or millions of it that's right approximation will change now well so it would always be this formula it's just that once you start aggregating across different so this the way you should think about this is you've brought your  
[00:59:34] economy down to the level where each of these js you can think of as a perfect substitute with the other js so if you haven't um if you haven't done if you haven't gone down to that level where things become homogeneous  
[00:59:47] then the first order approximation still works that's the first line but the second line is no longer true because in your numbers then you look down at what deep how deep do you go down you have thousands of so we break things  
[01:00:00] into bilateral so in the structural model that we have we break things into a sector um sector pair by destination origin so german chemicals industry buying russian  
[01:00:14] gas that would be then you assume essentially the perfect substitutes within the chemical industry even though or not no we don't assume perfect substitutes within the chemical industry we assume perfect substitutes within the  
[01:00:28] industry uh country pair so for example you're buying so germany is collectively producing a thing called um whatever is the chemical industry output or plastics or whatever that's a homogeneous output for us  
[01:00:43] and then when germany is sourcing let's say uh petroleum from different places we think of each of those as being an imperfect substitute for the alternative sources where you could source that energy that  
[01:00:57] makes sense can i jump in as well but marcus the nice thing about david's sufficient statistics formula is that exactly if you want it to go finer you can do it using the formula right no i see a different formula but i just  
[01:01:11] want to know in your calculations how deep how fine did you go no because if you don't go so deep then you assume more perfect subsequent ability and get you no yeah we there we did it at the industry level and we have 30 sectors in  
[01:01:24] germany and one of these to say the chemicals industry as a whole okay but if i cannot move gas because from one chemical industry plant to another one because of the pressure is not there  
[01:01:36] you would not capture that in a sense that's right yeah so so we do have so as i say for us the homogeneous good is country country sector specific and that's really a data  
[01:01:51] limitation it's just we're not able to drill any further that's the most disaggregated data that we're able to can you adjust your formula even without data say oh i put another cautious term in it just within this sector there's also  
[01:02:05] some friction yeah so so what i just want to say is i haven't mapped this to the data yet right the formula is the formula then the question is okay well are you mapping it to the data correctly and i think that's exactly that's  
[01:02:18] exactly the thing i want to highlight is the key sources of uncertainty in this formula because that's ultimately when you want to map things to data that's that's where you have to um make that connection so um the other thing is it's still a local  
[01:02:32] approximation or it's still estimating so where the estimate is coming from because if there are large shocks things might be very different that's right so if you so for the purposes of this talk i limited myself to a second order  
[01:02:47] approximation we could have gone further we could have done third fourth order but ultimately all these higher you're you're absolutely like but all the higher order effects are all gonna depend on how quickly these expenditure shares are changing so that's why i  
[01:03:02] think once you understand that then you can say to yourself okay well maybe i don't believe any given structural model so much but i can also look at historical experiences where there were big shocks and i can look at how much these shares actually changed in practice  
[01:03:16] so you could the opec 1970 exactly so that's what i'm about to do on the next slide okay okay great so so okay so the key sources of uncertainty are are the three things that i'm i'm going to talk about so the first one is just the shock to  
[01:03:30] the size of the shock which is the reduction in energy imports so that's the delta log m that's quantities the other big source of uncertainty is the change in the expenditure share on  
[01:03:44] energy and that term is where all these issues about complementarities the essentialness of these goods the nature of import substitution and the production network's robustness  
[01:03:58] they're all going to be inside that term so any model you write down will ultimately have an implication for how that number changes and that's the thing that's going to discipline how welfare changes in your model and then the next term that i think there's uncertainty if  
[01:04:11] you assume ces or do you assume no well in this formula i haven't assumed anything about the structural uh production functions this is completely non-parametric formula so the last term that we've got in here that i think is also important and  
[01:04:26] there's a lot of uncertainty about it there's a change in employment and these are principally going to be due to things like negative aggregate demand effects and ben is going to talk about these in more detail later but that's where you would plug that number and so really we have to just decide what these  
[01:04:39] three numbers are okay so and so these three things these three terms are what i'm going to be focused on uh so if you can go to the next slide okay so let's first start off by doing a back of the envelope kind of orders of magnitude  
[01:04:52] type of calculation so as ben mentioned we're going to assume that the reduction in gas in terms of quantities is about 30 percent that's the amount that we think where germany is getting from russia in  
[01:05:05] terms of natural gas that it's not able to substitute to other sources and as ben mentioned the expenditure share on gas as a share of gne and gdp is roughly about 1.2 and this is sort of  
[01:05:20] an annual number that um it won't vary hugely from year to year unless something uh dramatic has happened so how are we going to discipline the formula where our first starting point is to do exactly what marcus said which  
[01:05:34] is to look at uh the oil shock in the 70s and try to use that to discipline how much we think the expenditure share on or on energy might change in german so during the entire period of the 1970s  
[01:05:48] the expenditure share on oil increased at the world level from about two percent to um about eight percent so first after the first oil oil shock in the mid 70s it jumped from two percent  
[01:06:01] to about four percent and then after the iranian revolution in 79 it jumped again to eight percent and then it slowly came back down as we went into the 1980s so there was a quadrupling of the um of the uh expenditures on oil during  
[01:06:16] that decade if you kind of accumulate it all on and so we can say okay let's imagine just as a back of the envelope that we have a similar thing happen in germany where there's a quadrupling of expenditures um on on oil in response to  
[01:06:29] this shock and if you do that then you can just kind of plug and chug and see what you get at the other end so the first order effect is 1.2 percent times the log of 0.7 because this is a log approximation so log of 0.7 is roughly about  
[01:06:43] minus 0.3 like 30 and then on top of that you've got these non-linearities that come from complementarities which to add it on is going to be the change in the expenditure share which is going to go to 4.8 percent minus 1.2 percent that  
[01:06:58] 3.6 times a half because this comes from a second order or taylor expansion and then the size of the shop and if you add all of that up you get something that's close to minus one percent and so this sort of gives you a ballpark figure  
[01:07:12] of what you should expect from a model that's able to match what happened in the 70s in terms of the oil oil price sharks to go further and put and that's only it's only the price shock no it's not the redistribution of the energy within the country  
[01:07:26] is this fair to say uh sir what do you mean by that it's just i have oil supply coming from suddenly it costs three times as much and expenditure shares are shooting up but i'm still able to redistribute the whole thing efficiently within germany so when you  
[01:07:41] say i'm able to redistribute everything efficiently within germany um it has to be the the thing that has to be true is that the equilibrium has to be constrained efficient so what that means is if you have some kind of quantity adjustment cost that you have  
[01:07:56] to pay in order to shift resources or if um for example factors are in mobile across sectors or things like that those would be taken into account um what wouldn't that would kind of  
[01:08:10] influence your the final number you get for the change in the expenditure what we're not taking into account is something if you have um like genuine inefficiencies like market power or something like this where we're  
[01:08:24] abstracting from those kinds of frictions okay so this is the back of the envelope now to give it a little bit more color we can use specific structural models to basically try to fill in the numbers here so if we go to the next slide so  
[01:08:39] this table sort of shows you different ways of getting at the same object which is that change in the expenditure share which is going to be really key for thinking about how costly this is going to be the first column is just the sufficient statistics approach that i  
[01:08:52] showed you where we calibrated it using the oil price shock in the 70s and we got something on the order of about a one percent loss in gne which works out to to be about 400 per capita per year for uh  
[01:09:07] german consumers or german households then on the next three columns are using different structural models specific models with calibrated elasticities and so on and so forth so the  
[01:09:21] the last two columns are the ces model that then showed you which are kind of very simple models where you just pick an elasticity of substitution that's very small like point zero four and then you feed in the shock and you see what happens  
[01:09:34] and you can either do it to energy as a whole or to gas as a specific input inside that ces bundle the second column is a big structural model that comes from a paper that that i have  
[01:09:48] with emmanuel fari where we try to match exactly all the data on the world input output uh network so our model has 40 countries every country has  
[01:10:01] 30 sectors we match expenditure shares in terms of how countries and sectors are connected to one another we use elasticity estimates for the domestic economy that are very very low  
[01:10:15] so across sectors within the domestic economy across different kinds of consumption goods between value-added and materials we pick very low elasticities and then we pick trade elasticities that match what the literature has been estimating in terms of trade  
[01:10:29] elasticities and we just run that model just sort of uh if you like an off-the-shelf model with complementarities that matches the whole structure to see what kind of number do we get do we get something where we completely off somewhere and we find numbers that if anything are actually  
[01:10:43] smaller than what the standard ces calibration shows or the back of the envelope suggested and the reason there is because in the model actually that substitution effect the fact that there's lots of connections all over the place and everybody can adjust across  
[01:10:57] different margins actually winds up making the numbers if anything smaller rather than bigger um so david can i trust so far kirov would like to do an a different exercise he would say if you want to achieve a 30  
[01:11:12] percent reduction in energy or gas how much does the price have to go up did you do an exercise like this too yeah no we didn't do an exercise like this um  
[01:11:27] so in in the big model i mean so that it would come down to the effectively the uh elasticity of demand that you're assuming for gas i think the question has to be spelled  
[01:11:39] out very precisely because in these model it's there is no such thing as just sort of gas in in this model there's different sources of gas and different people feel differently because they're buying it  
[01:11:52] from different places but but we haven't done that that's certainly that's something that we can we can experiment with sorry sorry no no we did actually do this exercise in these simple  
[01:12:06] uh ces type models here in the in the end it's in the appendix and where we didn't do it is in your model david yeah that's right and and then it depends just on the elasticity of substitution um obviously  
[01:12:18] right so uh in particular uh with a very low lcd of substitution the price has to go up a lot there's an appendix figure um if you if you look at the appendix i put in the uh in the chat you you'll find it i i  
[01:12:32] think for the very very low elasticities which are probably too low where we want it to be really conservative we put a i think the the price has to go up by a factor of like nine or something like this yeah i think that's right factor  
[01:12:45] it's a huge price increase but the the point there is that we think it's actually you know unrealistic and too large why because the elasticities are so low and it comes back to this point that david made you know whatever you believe about the elasticities is going to have  
[01:13:00] implications for what's going to happen to the expenditure shares if you think the electricities are really low it's going to give you crazy movements in the expenditure chest and where do where do those come from well from the prices i guess that's right  
[01:13:13] i just want to say we have to speed up because we're running already 27 minutes so finish in about five minutes great so i just wrap up very quickly about what the model these numbers on the previous page are leaving so i think an important  
[01:13:28] thing that i've left we left out of that previous table are aggregate demand externalities and reductions in employment but i think it's helpful to kind of separate the physics from the macroeconomic stability part of this so i think those first two terms that had  
[01:13:42] to do with energy the reduction in energy and the change in the expenditure share on energy those are things that are determined by the physical nature of the production structure and then there's a different part which is the macroeconomic consequences in terms of employment and that's something that ben  
[01:13:57] is going to talk about briefly but it's also something that depends on the policy response in germany and the amount of physical support that there is nevertheless we tried to be as conservative as we could picking very low elasticities having the elasticities  
[01:14:11] rounding up our numbers putting in adjustment costs wherever we could and nevertheless we can't just it's it's implausible i think to generate numbers as big as for example 10 reductions in output and welfare and i'm going to stop and hand it back to them yeah and just  
[01:14:26] to be clear right the the headline number that we have so three percent gdp loss is essentially like the largest number here which is the simple ces model that doesn't have any substitution via trade um and then  
[01:14:41] we rounded up that two point two percent two three percent um exactly to have sort of a safety margin for uh these amplification effects and i'll talk about that a little bit uh in a second um one little qualification also  
[01:14:56] these numbers here they they have uh the substitution via trade um we do think those are probably a bit too optimistic um why because uh we think you know trade linkages and trade relationships  
[01:15:09] will take longer to adjust than just say six months um so you know probably these these numbers here or maybe something like this are the more relevant these numbers like whole 100 euros is it  
[01:15:22] a month or a year or is it an annual number it's yeah yes yeah we just took total growth uh gross national expenditure in germany and then uh say calculated 2.3 of that that comes  
[01:15:35] out to be 900 years okay um okay so because so you know when we wrote the paper then uh essentially a lot of people said you  
[01:15:50] know obviously you're leaving out some things um which uh some of these criticisms were uh fair others were not so much or just misunderstanding uh basically but um you know one that that's obviously  
[01:16:04] potentially important and that we did leave out in the numbers that david had just shown you is sort of a good old fashioned sort of changing aggregate demand amplification um so i wanted to briefly tell you about this and particularly there's a nice  
[01:16:17] paper by my co-author christian buyer with some students and postdocs i think i'm uh here at diw who wrote a nice sort of follow-up paper if you want about this okay um essentially what they took is  
[01:16:32] they took sort of a standard kanjin model in this case with heterogeneous households and then fed in sort of an aggregate supply shock that's meant to capture the type of shock we have in particular they assume that  
[01:16:46] tfp initially dropped by 2.2 percent why 2.2 because that's the the the number we have here and then they additionally put in a three percent capital obsolescence shock so kind of a this is a large uh a  
[01:16:59] negative shock here okay the other thing that's nice about this kind of model is you can think about inflation which is something that's important in the debate okay so here's what comes out of their model i'll go kind of fast  
[01:17:12] so essentially the the key figures to look at here is this one here a bip is german for gdp okay inflation is just inflation and essentially they get that you do get some amplification in particular from uh  
[01:17:27] 2.2 percent to a roughly three percent um so that's the aggregate demand uh amplification that's the the mechanism that you know everyone has in mind um so you know you you get the numbers are higher but  
[01:17:41] they're still sort of within this sort of safety margin we left ourselves by rounding up uh the numbers from 2.2 to 3 the other thing you can talk about now is inflation here um inflation in this  
[01:17:54] model uh shoots up by a little more than two percentage points what's really important here obviously is what monetary policy does um what's the assumption here the assumption is that monetary policy operates the taylor  
[01:18:07] rule um what does monetary policy do here therefore it sort of leans against the wind and uh rises it raises interest rates to sort of uh choke off uh some of the rising inflation that's probably an optimal policy response to a certain  
[01:18:21] extent uh uh because but when here the inflation goes up only by two percent is this for the the time period is years or is it quarters per month this is years i'm pretty sure  
[01:18:34] and but it goes up only by two percent and uh essentially you only increase the intercept at 25 basis points that's it uh no no the interest rate here goes by up by a hundred basis points right so it  
[01:18:47] goes from always from zero that's a change in the interest okay yes um yeah um let me don't follow our taylor principle at all essentially you don't keep up with inflation and it's assumed that the anchor never  
[01:19:02] breaks the inflation that's definitely assumed that the anchor doesn't break um no no you do follow a taylor principle it's just that uh that there's gdp in there as well and  
[01:19:16] gdp is kind of uh uh for some time okay okay let me um speed up and sort of work towards the conclusion and just want to briefly give you a sense of other studies so you know at this point it's  
[01:19:29] not the case that our study is the only study um or this together with this paper by christian and co-authors at this point there's a lot of different studies with lots of different um methods some much more sort of structural macro models  
[01:19:44] some uh very different uh methods the there's a very nice overview of the current estimates of this in particular uh uh but in a report by the german uh council of economic experts  
[01:19:58] and here's sort of this survey table they have and what i just wanted to point you to is uh this this column over here um which has this sort of gdp reduction from different so these are slightly different scenarios um but the  
[01:20:13] point here is essentially all these numbers here are something like you know uh two percent or one and a half percent so none of these numbers are like five percent or ten percent okay uh there's also some inflation numbers here and you can see they're like maybe two and a  
[01:20:27] half percent and so on more generally this is just sort of a brief reading recommendation this uh economic council of economic expert reporting i just think is really very well done i really recommend it um they uh uh have both a  
[01:20:41] german version and sort of a shortened english version um they're you know obviously them being the council of economic experts they're you know quite guarded and put on lots of uh you know caveats and and qualifiers which obviously is the right thing to do um in  
[01:20:55] the text they don't have a bottom line number they come up with um but uh faulker wieland who is one of the sages i guess they they're always called in germany um gave a a a press conference where he essentially  
[01:21:09] said you know this is means we think it would be three to five percent gdp loss of a full cold turkey import stop okay i'm just briefly again i wanted to say you know this is very well done and and they really have a very good team of uh  
[01:21:22] of economists there uh and uh the whole three to five percent for how long for permanent loss forever that uh no i think this is just the first year um i don't know what the assumption there is uh how long it's  
[01:21:36] gonna last uh this is for over the first year yeah um okay just briefly if there's time um i wanted to just uh talk about  
[01:21:48] some criticisms that we've received um uh that we haven't discussed yet i'll go fast with this one tom krebs wrote a a blog post i guess or a newspaper a little article and he essentially uh didn't like the fact that we have in  
[01:22:02] our computational model um uh which was the second column in in the in the numbers that david showed that we have the same elasticity of substitution in different industries so he thinks we should have a separate elasticity of substitution for the  
[01:22:17] chemical industry and uh that should be lower than the number we currently have that's 0.05 um i think that's a fair point you know in fact what you can use there very nicely is the uh sufficient statistics  
[01:22:31] formula that david just showed from the kaifari model uh we need to think harder about to do this but again it's going to come down what you think is going to happen to these expenditure shares and we think unless you think something crazy is going to happen there uh you know you're not going to get numbers  
[01:22:45] that are bigger than uh three percent okay um here we also got some uh criticism um by i guess the the chancellor and uh the economics minister these are sort of uh  
[01:22:57] signed referee reports if you want um uh so so they they said these kind of things here um uh you know we're forgetting sort of where is the gas actually supposed to run through where are the pipelines what is the  
[01:23:10] regasification capacity or how big said sheer physics stands in the way of these macroeconomic models the time it takes to build these pipes and so on um i think these are uh kind of good uh criticisms uh  
[01:23:25] i i think a large part of this is not necessarily about our macro models you know i do think it's important to emphasize that macro models do respect physics you know what's more physical  
[01:23:38] than resource constraints and and production functions which we obviously have i think the way to read this is one of three things too that are listed here and one that marcus made me think of which is um uh you know maybe what they mean is that  
[01:23:53] the gas struck shouldn't really be thirty percent um uh instead it should be somewhere closer to fifty five percent because maybe we can't uh you know substitute we can't get this extra gas from norway and so on so quickly maybe we're too optimistic there maybe  
[01:24:06] we should think hard about that um then the other thing uh that's let me actually say that first um is maybe a good way of thinking about it is what marcus said maybe what they want to say is that we need sort of finer sectors and there's sort of a chemical industry  
[01:24:20] in one place and a chemical industry in the other place you need to move things around again what's kind of good is we can use david's sufficient statistics formula i like marcus's comment that we can put in some sort of a safety term there under some assumptions so that's a good idea  
[01:24:34] um another way to read this is you know maybe what they want is some sort of a spatial model with transport costs um as in like esteban and steve's work um and so that that's something interesting to think about  
[01:24:49] okay we have to come to an end pretty soon actually because we're running already half an hour over oh sorry uh okay one or two words about france and other eu countries and  
[01:25:01] um so you know we did this we wrote this report here for the french council of economic adviser um with camille landay and philippe martin where we essentially used um  
[01:25:15] david's computational model and just produced numbers for a bunch of different countries the exercise we did there is two exercises the first one is more similar to what we did in the germany paper which is  
[01:25:30] just introduce trade barriers that basically choke off a trade between the eu and russia okay sort of a a complete import stop and  
[01:25:44] then you get some gdp losses and uh these numbers here the the blue ones um that are labeled embargo and the interesting or potentially problematic thing you get there is that there's  
[01:25:58] large heterogeneity in the economic costs across countries in particular there's these eastern european countries like lithuania bulgaria so essentially countries that are very integrated with  
[01:26:10] russia that would have very large um losses in terms of the economic losses we the other things is among the european union this is sos agreement to  
[01:26:24] bring gas across the union essentially right so they're in the model so so in the model we let prices do their work so you know the gas then gets reallocated uh through the price mechanism in the model  
[01:26:38] so i guess there's an interesting question you know to what extent that would replicate what that that agreement would uh give you um but here the main point really that i wanted to just make here is we did another simulation which is what happens  
[01:26:52] if instead of doing full embargo you do a tariff okay turns out that a 40 tariff in this model at least is actually sufficient to completely choke off um uh  
[01:27:05] all energy imports um from russia for the big majority of uh these countries uh uh here uh to the to the eu okay um you can see the the tariffs here are the red diamonds one thing that's kind  
[01:27:18] of nice about the tariff is that the costs are much more equally distributed across countries and sort of smaller on average which i guess chimes with people's intuitions okay so let me conclude um you know our conclusion or or our message is uh  
[01:27:33] the cost of an embargo or an import stop uh would be substantial you know three percent of gdp is obviously not nothing but not catastrophic again ballpark somewhat smaller than uh covet we think these are conservative  
[01:27:47] estimates we really like try to you know make all the llc's elasticity super low and the numbers we that generate the three percent have like basically no substitution via imports and we rounded them up to allow for  
[01:28:02] these amplification effects um policy let me just say two words and then conclude um you know i'll pick the important ones um you know i do think it's important uh to  
[01:28:15] let the price mechanisms work and you know we want these prices of gas and oil to go up to a certain extent and we want uh people to substitute obviously this comes with political economy problems but you know just from a purely economic point of  
[01:28:29] view i think this is important uh you don't want to have rationing plans that somehow let the shock fall entirely on industry or households we did some calculations then it actually can potentially be much worse monetary policy should raise interest rates to  
[01:28:43] control inflation you don't want to tax a subsidy petrol say you know again given the size of the shocks the economic costs we think a country  
[01:28:57] like germany with relatively modest debt-to-gdp ratios could potentially uh you know use policies to alleviate some of these losses  
[01:29:10] in particular avoid financial spillovers effects what are the policy tools well kind of exactly the same ones we've used during kovitz so furlough so could survive in in germany you may have to bail out certain companies so for example for the  
[01:29:25] chemicals industry i mean it would there's a chance it could get really bad in which case uh maybe you have to bail out bisf for example um okay okay so let's uh come to a close  
[01:29:39] thanks a lot uh to to you ben and david for outlining uh your estimation procedure and the way to approach this thing so we learned a lot and i think we got a much finer detail now and as uh probably i would argue that there has a  
[01:29:54] lot of value to do this quantitative macro models but it also has a lot of value to do some other studies and then as we put everything together we get a richer picture how things play out uh hopefully at the end we hope that the war will be  
[01:30:06] over soon in ukraine as people suffer and uh that's quite a tremendous hardship and for all of you who were hanging out till the end thanks for hanging out till now and um just the quick advertising for next week  
[01:30:20] we have elena kipakova with us she will talk about other aspects of russian sanctions probably more on the financial side and how things really play out and how the payments are done and how you can get around it and why the hoople is still so  
[01:30:34] strong and all the other aspects and i hope to see you next week and thanks again to david and to ben um take care cheers