How Big (Small?) are Fiscal Multipliers?

Ethan Ilzetzki, Enrique G. Mendoza and Carlos A. Végh provide new empirical evidence (using a different method to the recent IMF study). The paper is here and the abstract is below.

Abstract: We contribute to the intense debate on the real effects of fiscal stimuli by showing that the impact of government expenditure shocks depends crucially on key country characteristics, such as the level of development, exchange rate regime, openness to trade, and public indebtedness. Based on a novel quarterly dataset of government expenditure in 44 countries, we find that (i) the output effect of an increase in government consumption is larger in industrial than in developing countries, (ii) the fiscal multiplier is relatively large in economies operating under predetermined exchange rate but zero in economies operating under flexible exchange rates; (iii) fiscal multipliers in open economies are lower than in closed economies and (iv) fiscal multipliers in high-debt countries are also zero.

28 thoughts on “How Big (Small?) are Fiscal Multipliers?”

  1. Welcome to a highly indebted rich country… thanks for the light, Mr. Lane, let’s see what heat comes!

    Have we ever figured out if we are a developing country or not?

  2. Another SVAR paper without any substantial “financial sector” data. I thought that even central bank economists are finally now considering macro models with an explicit financial sector!

    Another point, how does these models behave when faced with structural changes – one that springs to mind is Ireland joining a monetary union to become a region of euro area with no (or more limited) monetary policy.

  3. This paper further reinforces my view that economists simply flirt shamelessly with mathematics, and have no real understanding or insight into the subject or indeed its application to their field.

    This paper examines an algebraic linear model of the economy in relation to government spending and then proceeds to draw conclusions from there. It is difficult to describe how utterly bankrupt any such analysis is in lay terms, but in an attempt let me make the following analogy.

    Attempting to analyse something as complex as an economy with a linear model is like attempting to analyse a hurricane with a ruler. To be sure it may be a very fine ruler, and you may have an entire warehouse of them at your disposal, but at the end of the day it’s not a very legitimate exercise and is unlikely to lead to any useful understanding of the hurricane.

    You can tell the authors are out of their mathematical depth from statements like these

    We assume that the matrices A, B, and Ck are invariant across time and countries.

    Economists; I am disappoint. I thought your ability to browbeat with mathematics was far more honed than this. And the system they use is even discrete. Surely after all these years you’d have learned how to lash together a system of ordinary differential equations. As with every other discipline in finance, economics it seems is more about bluff and bluster than serious analysis.

    In conclusion, this paper isn’t worth very much and no serious conclusions can be drawn from its analysis. I’m quite serious.

  4. OMF
    I think your confusing mathematics with statistics with economics with reality. Its nice that you found the forum…but your coming across, in every single post, as a math-troll. Try this – treat the site as a research site. Critique, by all means, but then suggest some ways of moving things forward. Frinstance, how would you propose to fix a set of ode/pdes to the frequency of the data? That would be very useful… Sneering, jeering and so on is troll like behaviour. Try acting like an adult.

  5. It is quite clear to me that the core of goverment fiscal strategy should be geared towards capital projects that reduce our transport fuel costs dramatically.
    While high fuel consumption perversely adds to GNP and the consequent tax base this obviously does not add anything to our standard of living if those BTUs are expended on services that are wasted.
    The core of the Western financial crisis is the undervaluing of real capital stocks and depletion entropy.
    Since 71 the last tie with the best symbolic representation of capital has been broken and depletion econimics became the only show in town.
    Therefore the major drive in energy was towards effeciencey of existing stocks of capital rather then energy capital creation – while the Irish goverment cannot change global monetory policey it can resist its excesses by engaging in a large strategic drive to increase Irish capital resources.
    This would entail the continuation of the Luas system (A line from Mahon to Ballincollig in Cork – my favourite capital creation idea would transform the city and have much greater potential then the newest Luas line given the relatively high density of people living close to the lee which therefore has a Linear nature)
    People who argue against such Ideas on the grounds of fiscal prudence are ignorant of the wealth creation vortice that these projects create which is Independent of the ventures booked profits or losses given the above nature of the flawed monetory regime we live under that provides the same cost of capital for both capital increasing and capital decreasing consumption.
    Therefore the flawed ventures which have artifical savings such as a bus service following the above route that has very low capital cost but much higher running costs per person need to be abandoned as these artificially more effiencent solutions are a artifact of a distorted interest rate environment which unfortunately the ECB is attacking us with.
    Their agenda seems to be to reduce and destroy the periphery capital stock over time to defend their currency against Gold – this is similar to Volcker’s strategy in the early 80s which unfortunately has devastating effects on Americas Industry and Utilities and merely forced manufacturers elsewhere due to the high interest needed to build capital in the States.

  6. @OMF, Brian’s right. If you’re so smart, have a pop at modeling this a little yourself, I’ll be more than interested to read your work. Simply saying ‘this isn’t worth very much’ is not in and of itself a critique. Thoughtful critiques are always welcome on this site I’ve found.

    Moving swiftly on, you’ve got to respect the authors for being up front about the dodgy nature of the data they are dealing with, but let me take issue for a second with the sVAR methodology. All of the previous studies mentioned in the previous important posts on multipliers seemed to use xVAR-type estimations. VAR is a pretty poor way to estimate any specific features of the economy. It suppresses information and its results tell you nothing about how the economy works.

    VAR is a further modification of modifications of Box Jenkins, and is intended to be an alternative to simultaneous equations; so, of course, the first problem is getting a stationary series. Stationary? If we want to look at the patterns of fluctuations in a given system, we want stationary data. But the whole point of studying growth is to determine the trend, because we think that what determines the trend also determines changes in the other proportions and aspects of the system. And crisis? In a crisis the positive trend has been replaced by a negative trend – that is the change that we want to study, and we want to examine its implications for the basic relations of the macro system. Differencing (or any other way of detrending) removes a lot of the information we want.

    Turning to VAR itself, remember, that Sims introduced the approach as a modification of structural econometric time series, itself a modification of ARIMA, in reaction to arguments over the unhelpfulness of economic theory, not surprising since must of neo-Classical theory really isn’t any help. So Sims basically removed not only all restrictions derived from theory, but all restrictions altogether. VAR treats all variables as endogenous; all variables depend on lagged values of themselves and on lagged values of all other variables. We write a vector of the variables and regress it on itself, plus an error vector. Then by substitution we eliminate the succession of lagged values, and write the vector as a function of the lagged values of the error vector, in orthogonal form. There is a good deal of debate as to how the resulting equations should be interpreted, but what is undeniable is that the procedure has discarded or buried a lot of information in order to reach a result that is far from transparent.

    Suggestions: First, just distrusting the simultaneous equations approach doesn’t mean you have to go all the way to VAR. You could do an error correction model. Establish the level of income you might expect theoretically to hit as a result of multiplier processes, then define the difference between that and actual income each period (month/quarter/etc), and set up an equation tracing the impact of stimulus spending on this difference, to see if it diminishes. Or, since we don’t want to lose information by making the series stationary, you could try co-integration, for example, with government spending and aggregate income/output adjusted for the effects of changes in investment and exports. I’m not sure if anyone has done this for the Irish economy, because I’ve not looked yet. If anyone knows of any literature out there on this, please drop it into the comments.

    Perhaps best of all, why not build up a macro model based on household income spending equations and productivity equations, using very current data? Analogous, for example, to the Fair Model of the US. (These equations could also be establishing using cointegration procedures, which could serve as a check on other more simple-minded regression methods).

    The advantage of this is that you can test to see if there are changes in private saving, and / or debt servicing or in spending on imports. If government spending rises – say the government hires workers to finish NAMA’d construction projects – taxation the same – but the data shows that there is not much in the way of repending, so there is very little further increase in employment, it must be because of a rise in household saving, or of debt servicing, or a shift to imports. This is not a matter of theory, something that Sims could complain about; it is a matter of national income accounting, pretty basic arithmetic. This kind of thing should be discoverable and testable in the data. It could be confirmed by interview data or other field work. And the model could then be adjusted. By graduate students 🙂

  7. I have been asking questions about the unseen, inner workings of an economy which gives traction to these Multiplier things. I gleaned a few drops of enlightenment in the paper: different economies (with different attributes) display different outcomes. If you run this out to the end of the piece of string – then each economy is unique (or almost so), hence it should have its own unique signature Multiplier (or set thereof)? What about intertemporal effects? Does this mean you need to re-calc the Multiplier for a specific set of conditions, and if these vary you have to re-re-calc? Log tables and slide-rulers at the ready lads! I think you may need then soon (see below).

    I’m going to leave this matter alone. It’s got my neurons in a tizzy.

    Anyone check out the latest on the Great Mortgage Fraud in the US? Looks like it will be a dilly!

    Brian P

  8. Not sure about the validity of this.
    The effects of government spending are influenced by a myriad of factors which have not been controlled for here at all.
    First off – there is a difference between a government “stimulus” package and a routine increase in spending. The psychological impact of the first is much greater than the second
    Second the only way of truly stating the effect of the multiplier would be to have two identical economies in identical times and have one increase its government’s spending and the other reduce it. Otherwise there are too many confounders.
    This is really a poor observational study dressed up as science. I suppose it’s the best that can be done though.
    Maybe they have done this (and if they have I’m sorry for missing it) but they should insert variables such as business confidence prior to the increase into their equation.
    The interplay between these indicators and the multiplier effect would be interesting.

  9. Apart from the fact that we’ve had multiple discussions of multipliers over the year, this specific paper was already mentioned in September.

    See this thread

    There are also a number of other papers on the topic of multipliers from a thread in April.


    That last one is from IMF staff, although it carries a specific disclaimer that it doesn’t represent the IMF.

    IIRC the message for Ireland from most of these documents is that fiscal multipliers for a small open economy are going to be low and possibly even negative.

    Dr. Lane’s own paper from the ESR was also quoted in the Sept thread.

  10. If I am coming across as caustic, it’s because I’m feeling very caustic towards economists as a profession. Most economists were cheerleaders for the boom and now that the bust is upon us all I can see is rather empty studies employing the usual fig-leaf of mathematics to sell one ideological argument or the other.

    I’ve read several economics papers lately and a common trend in most is to employ some linear model or the other in some haphazard fashion. Regarding this, I stand by my comment about the ruler and the hurricane. As I see it, mathematics is mis-employed in countless economics papers, as far as I can see in an effort to give a veneer of scientific rigor to what are inherently subjective proceedings. My snarkiness aside, this is a very serious problem with economics as an academic subject.

    This study suggests that governments cannot Keynesianly spend themselves out of recessions. Now, while this conclusion may be valid, the method used to reach it—misapplication of a linear model—puts serious doubts on the credibility of the study. Had they simply left the model out, I for one would be more inclined to accept their arguments.

    Economists are supposed to be offering analysis of the economy and proposing ways for the country to extricate itself from its current position. But what’s being provided instead are very suspect studies, and dozens of conflicting opinions. And the alternative is the likes of McWilliams endlessly on the MC circuit coining a new term every second week. It’s very frustrating to think that governments may well act (has acted!) on the advice of someone picked seemingly at random out of this group.

    Having diverged for long enough, I’ll finish by saying that I personally regard government stimulus as something which can work well, but which probably won’t work for Ireland because we were too much in debt when the crisis began. If the government had the money(and there wasn’t a glut of buildings), a few capital projects would be money well spent. The government hasn’t the money, so there’s no point really discussing it.

  11. @OMF
    The goverment has no money because the static or declining money supply is being drained via mortgage payments to pay external bond holders including the ECB.
    You cannot increase tax revenues on a contracting money supply and hope for sunshine and daiseys.
    They are deliberately imploding the economy to extract the maximum amount of interest before collapse – where we will have to sell our equity on the cheap to survive day to day.
    They sell a thatcherite adgenda to a populace who believe it might work because it worked before little releasing that there was no austerity in the 80s ,just a transfer of wealth under a rapidly increasing money supply.

    Ireland is a econimic battlefield where the FED and the ECB can fight a limited war of attrition with gentlemen’s rules.
    Unfortunately these generals do not care much for collateral damage – we are all gooks in Paddy fields to them.

  12. I used to be called Paddy Fields at school!

    OMF and KC.
    Yes, much of what you say is correct. Sadly. Economists are most successful when they are free of enslavement to others and those on this site are freer than most, but von Mises and Ken Rockwell are both very worthwhile reading. History trumps theory. We know what hasn happened before and we know therefore, what is likely to happen now. We erect laws and instiututions which are meant to objectively govern us. When those who design them are flawed, the products reflect those flaws.

    Most Fabians are economists. One World government requires work and sacrifice. The rest of the world should not have its capital stolen. By destroying the capital theft machine, economists have done a good days work, albeit at the cost of some wealth from the feckless and greedy. Learn what is real if you want to avoid disappointment. Human beings are challenging their conception of magic every day, but those who refuse to learn in one way will learn in another!

    Take a look around and realize that the world is altering faster than ever before. Do you want to play a part in that? Then organize!

    US economists are all controlled by Fed thinking. So easy therefore to destroy their finance system! There are factions and those who are trying to oppose equality of opportunity but they are very clumsy, often working against their own aims. To get an idea of how far this goes, consider the Russia experiment? From 1900 to 2000. It is now perhaps the most christian of states. Obviously, I am insane and these are merely ramblings of a broken mind. But like a broken mirror, some reflections are valid?

    What a hoot, sneering? Yah!

  13. This paper further reinforces my view that economists simply flirt shamelessly with mathematics, and have no real understanding or insight into the subject or indeed its application to their field.

    I tried to differentiate this sentence and set it equal to zero, but it still doesn’t seem optimal.

  14. @omf

    I hope the broad brush you paint economists with isn’t the one you work out your maths with.
    you have a great thesis there…..
    An important one
    Go go

  15. @ PD: “Obviously, I am insane and these are merely ramblings of a broken mind. But like a broken mirror, some reflections are valid?”

    “Its all realtive” – as Mr Einstein said! If your reflective surface is plane, then your image is virtual and inverted. So! – now you know! Your a virtual, inverted … …! Thanks for the laugh!

    @Eureka: “Second the only way of truly stating the effect of the multiplier would be to have two identical economies in identical times and have one increase its government’s spending and the other reduce it. Otherwise there are too many confounders.”

    My sentiments exactly. Hence my impatience with the abstract math models. They are after the fact – no predictive power whatsoever.

    Have you checked the state of play in the US over the Great Mortgage Scandal? If this massive, systemic financial fraud is confirmed – then the global financial crisis just got nuclear.

    Brian P

  16. @hogan
    On your question of whether we’re a developing country or not for the purposes of the Ilzetzki, Mendoza and Végh paper…IIRC the authors put us outside that category.

  17. My math background is from physics & I’d agree that the link Karl posted is accurate 🙂

    The accuracy of any mathematical model depends on two areas:
    -Is the logical reasoning sound & the model built correctly?
    -Have the constants in the model been estimated correctly?

    It is likely that economic modelling is complex and require many variables to capture the complexity.
    One variable -> One constant to be estimated.
    Many variables -> Many estimations.

    In mathematical modelling in physics or of chemical reactions on molecular level it is possible to repeat experiments and it is possible to vary experiments. Therefore it is possible to correctly estimate constants.

    On the other hand, mathematical modelling of economics…. Controlled experiments are not possible -> A model that is modelling the past perfectly might be possible. Extrapolating it into the future demand that no constants or relationships change. I find that plausible in a short time-frame, extrapolating it over a longer period & the predicted result might be accurate but I wouldn’t bet my money on it. If I was asked to do so, my question back would be: Are you betting your house and home on this or are you just willing to risk mine?

    My opinion is that the current economic models might be accurate but only in the extreme short term and I seriously doubt their accuracy over multiple discrete events (for example budgets; own, major trading partners & major employers).

  18. would it be fair to say that lots of multipliers are now under question? Not just fiscal, but monetary as well. At least two references come to mind.

    I think it is well worth trying to find out if any resultant multiplier has a general output effect, if so it can become a vital part of policy, equally, if it doesn’t then we should end the charade. Knowing is key.

    The concern I would harbour is that if fiscal and monetary multipliers don’t exist then it seriously narrows the spectrum of options available to any policy maker, so what other policy do you use?!

  19. Actually, the big question economists have to answer is this:
    After immense progress in all sorts of disciplines since WWII, how it is that their voodoo is on the brink of reducing the world to penury, strife and fabulously rich bankers?

  20. Richard, blaming economics for the financial crisis is a bit like blaming sociology for persistent income inequality.

  21. @Enda H
    “Richard, blaming economics for the financial crisis is a bit like blaming sociology for persistent income inequality.”
    Oh I don’t know. If some sociologists had claimed to solve income inequality, what would you think if it was proved they had not? All get tarred with the one brush.

  22. @hoganmahew,
    Good point, well taken. I agree that my analogy is not a perfect fit since there were problems in economics (e.g. failure to appreciate the importance of financial markets) that didn’t help this problem. But stating that economics “is on the brink of reducing the world to penury, strife” makes economics the cause. I don’t think economics is cause. Economics could have (cudda, wudda, shudda) helped put out the fire more, but it’s not the cause.

  23. @Enda H
    Yep, that’s fair enough. It is back to the dog that not only didn’t bark, but showed the burglar where the keys where (thus avoiding him having to break a window)… or maybe that’s accountancy… 😕

Comments are closed.