Why Is Deflation Scary?

by: Cynicus Economicus

I have several times in my blog questioned why it is that deflation might be a bad thing, including in my discussion of a new form of money. Having presented my arguments, I thought it might be useful to find out exactly why economists insist that deflation is something to be feared. Having read through some academic literature, I found an excellent paper which clearly shows that the deflation scare is exactly that - a scare.

There is absolutely no evidence that deflation will cause depression. I have therefore written a summary of the paper in question below, which is by Gregor Smith (reference at end). It might be noted that his paper is written in the polite tones of academic discussion, but the message is clear. He divides up his paper into some basic arguments and I will follow the format.

1. Deflation is associated with depression

Smith reviews the study of Atkeson and Kehoe (2004) who considered the empirical evidence for a link between the deflation and depression. He summarises their work by concluding that only ‘for 1929-34 is there a positive relationship between the inflation rate and the output growth rate’ and that ‘excluding 1929-34, there is virtually no link [between deflation and depression], even though there were other periods of deflation, especially under the gold standard’ (p1046).

My comment: The important point here is that there have been many, many deflations in which there has been a growth in output.

2. Unexpected deflations are associated with depression

In a contrast to the work of Atkeson and Kehoe, Smith reviews the work of other researchers who have utilised different analytical models which have found a relationship between depression and deflation, for example in Canada 1870-96. However, when models take into account other shocks in the economy, the correlations found appear to be coincidental rather than causal, and such shocks also serve to explain the only correlation found by Atkeson and Kehoe.

My Comment: Again, it appears that there is no evidence that deflation and depression are connected and as Atkeson and Kehoe identify, there are many examples that directly contradict the idea that deflation and depression are linked.

3. Sticky nominal wages

In his review of the literature, Smith finds some problematic findings on the relationship between wages and deflation, such that each of the studies fails to fully explain the relationship between wages and deflation. For example, he reviews the work of Bordo, Erceg and Evans (2000) which explains wage stickiness for the period of the early 1930s, but fails to explain wages in the slow recovery for the period of monetary growth after 1933. His conclusion is that there is a need for further research if there is to be a meaningful debate on how wages might be determined in deflations. In summary, there is a general lack of research that might present any firm conclusions on the relationship between deflation and wages.

My Comment: It is worth noting that, if wages were to remain static in monetary unit terms during a period of steady deflation, the recipient of the wages would find the purchasing power of their wage increasing. As such, the person would, in real terms, see an increase in their wealth. Even if the person’s wage were to decrease in a period of deflation provided that the decrease is less than the rate of deflation, they would still be seeing an increase in their wealth. Why such outcome might be viewed as problematic is entirely unclear. Actual wages in monetary unit terms are irrelevant without relating them to what they might purchase in terms of goods and services.

4. Debt deflation

Smith identifies Fisher's (1933) work as being the key work on debt deflation, whose theory Smith summarises as ‘depressions begin with debt liquidation, leading to deflation and then to bankruptcies, and to a fall in output and employment’ and that there is an association between these events and ‘a nominal fall in interest rates and a rise in real interest rates’ (p1050). Smith then identifies that Fisher is rarely cited in empirical studies, that the majority of studies cover the period of the Great Depression, and that the studies that cite Fisher provide ‘little empirical evidence on the mechanisms Fisher outlined’ (p1051). From this starting point, Smith reviews many studies, and concludes that ‘the historical research does not seem to me to provide much evidence on the debt deflation mechanism in the 1930s’, and that no recent studies are identifying debt deflation in a deflationary environment that might support a ‘spectre’ of deflation hypothesis.

5. Deferred Spending

The idea that deferred spending is a result of deflation is described by Smith as the worst argument made against deflation, noting that the saving from deferred spending will lower the rate of interest and increase investment. Smith is quite right to question the principles of deferred spending, but his questioning might be taken further.

Smith later looks at some modern deflations, such as Hong Kong coming out of the Asian crisis, and finds that there is a correlation in these cases with unemployment. However, in some of the cases that he examines, the deflation appears to be resultant from shocks (i.e. the shock of the Asian crisis itself for Hong Kong and the collapse of property prices, the property and equity melt down in Japan). In two cases, Britain and Canada in the 1920s and 30s, the explanations are less clear, though some of the deflationary episodes followed episodes of high inflation, and other parts of his analysis again cover the Great Depression. Without more detail on the episodes in these countries, with which I am not familiar, I am unable to comment on these.

It should be noted here that he is presenting a very limited number of examples in which there is a correlation between unemployment and deflation, and his conclusion is that, whilst he has found a link between deflation and depression (in the form of unemployment), he can not square this with the other studies. Most importantly, he points out that there are 'different kinds of deflations', meaning that there is no reason to link deflation to depression. His own examples, are therefore not indicative of the idea that deflation will cause a depression, or that deflation must be accompanied by depression. Even Smith, despite a genuine intention to examine deflation from a firmly empirical point of view, appears to have a basic confusion in the cases that he studies. The confusion arises from trying to impose economic theory on to the evidence.

The source of the unemployment in Japan and Hong Kong was, in both cases, the result of unwinding bubbles in which resources were allocated in unsustainable ways. The bubbles in the economies led to the destruction and misallocation of capital, and labour being directed into unsustainable businesses. Unemployment is inevitable. It will take a considerable amount of time for the bubble based businesses to unwind, and further time for retraining and redeployment/retraining of workers who are laid off as a result. It is a simple and logical explanation. Why on earth would deflation result in unemployment? As Smith himself identifies, there is a lack of evidence for this mechanism.

The interesting thing about Smith's work is that it is a review from academia (interestingly including work of Ben Bernanke). As was mentioned at the start, he uses polite academic language, which is the language of polite questioning and requests for further study. Nevertheless, he presents a compelling demolition of the idea that there is any evidence that deflation will lead to depression, or that depression accompanies deflation.

The most important point in the paper by Smith is that it shows that deflation can take place during periods of growth in output, that the deflations associated with depressions are mostly the result of shocks and were not causal. At the moment the world economy is undergoing a shock. That this shock might cause deflation does not make deflation a 'bad thing'. It would simply means that it is associated with a 'bad thing'. However, the scare of deflation has been used to justify the printing of money, low interest rates and the fiscal stimuli.

As I argued in an earlier post, it is not deflation that is a problem, but the move from inflation to deflation, or even high inflation to low inflation. What we are seeing in the stimuli and money printing is an attempt to prevent such an occurrence. For any borrowing undertaken at a fixed interest rate before the change, these result in real increases in the burden of debt for the period of the fix. Preventing this problem might be seen as a justification for the policy, but the broader cost of these policies is a risk of hyper-inflation. The scale of the potential problem of the increase in the debt burden has never been spelt out, but it is the only legitimate reason that might be used to justify the deflation scare. More to the point, the transition from high inflation to low inflation has an identical effect upon debt burdens (see notes for further explanation), but we have never heard arguments against moving from high to low inflation. Why is that?

At the moment, the depression is already taking place. That depression might create deflation is not to say that the deflation is itself problematic. When looking at the deflation scare, it is a genuine puzzle that the scare has been allowed to gain so much traction. There is simply no evidence that a deflation would take the economy deeper into depression. Despite this, all over the OECD there is a huge experiment in monetary expansion, and an explosion of debt, with the fight against deflation as one of the explanations for this policy.

It just does not add up.


Atkeson, A. and P. J. Kehoe (2004), "Deflation and depression: is there an empirical link?," American Economic Review, 99-103.

Bordo, M. D., C. J. Erceg, and C. L. Evans (2000), "Money, sticky wages, and the Great Depression," American Economic Review, 1447-63.

Fisher, I. (1933), "The debt-deflation theory of great depressions," Econometrica: Journal of the Econometric Society, 337-57.


Note1: I have included a previous discussion of deflation from a previous post in the notes below, and have added an additional example that shows that consumers do not defer spending:

In the following examples, it will be shown that the reality is that people do not delay purchases in the expectation of lower prices.

Example 1 – Fast Moving Consumer Goods

If shampoo manufacturers were to improve their output by 5% through a manufacturing innovation each year, their output of shampoo would increase, and this would reflect in a decrease in the price of shampoo. In other words, there will be a steady and continued deflation in the price of shampoo. According to the idea that consumers will delay purchases in an environment of deflation, in such a situation, consumers would choose to walk around with greasy hair, never buying shampoo in the expectation of further price decreases. Such a proposition is fatuous.

Example 2 – Hedonic Goods

Over the last few years countries such as the UK have seen the emergence of many discount airlines, such as Easyjet. The emergence of these kinds of airlines, and the increase in competition within the sector, has seen the price of air travel deflating. Much of the utilisation of these airlines has been by consumers using the discount airlines to have cheap foreign holidays, and this can be described as a hedonic good. It is an entirely discretionary expenditure as there is no necessity to go on holiday to another country. Despite the continual deflation, there have been many years of continual expansion in the discount air travel market. The deflation has not prevented consumers from taking flights to go on holiday, but rather has had the opposite effect.

Example 3 – Computers

Personal computers (PCs) are an interesting case, as they have year on year improved performance and year on year seen deflation of actual prices. It is also an example that includes both business purchases and consumer purchases. Despite the ongoing deflation in the prices, the market for PCs has had a long period of explosive growth throughout this deflationary period. It seems that the steady deflation in prices has had no impact through the postponement of purchases.

Example 4 – Special Cases

Remaining with the example of PCs, it is possible to construct a hypothetical example of how consumers might indeed delay their purchase in expectation of deflation. If one of the large computer manufacturers were to announce that they would be introducing a new type of computer in the coming year, and that the computer was to offer twice the performance at half the cost, it is quite likely, assuming their claim were credible, that consumers might delay their purchase of computers in expectation of this future deflation.

Example 5 – Purchasing on Credit

The example of purchases on credit with interest of goods and services suggests that there is a fundamental flaw in the deferred purchase argument; that theorists have misunderstood the psychology of consumers. Purchasing a product on credit at interest is a real increase in the cost that will be paid for the good, whilst saving the money with interest paid is a real decrease in the cost paid for the good. Despite this, many consumers do not defer the purchase, but instead choose to purchase the good at greater cost now, than the cost in the future. Furthermore, in sectors such as computers, the deflation of the good does not prevent purchases utilising credit.

If the thinking of those who argue against deflation is considered, such a deflation is a ‘bad thing’ as consumers withhold their money in expectation of lower prices. If this logic is followed, then the new and more effective design of computer is not a good thing for the economy, as it has created a deflation in the price of computers, and has caused a delay in the purchasing of computers. However, once the computer is introduced, it will make more computing power available to more people. How this might be a ‘bad thing’ is not entirely clear. Everyone who purchases a computer sees their wealth increase, as they are able to enjoy relatively more computing power in relation to their income. They are quite literally wealthier.