Why Not Deflation?

In "Inflation hurts; so why not deflation?" Rodger Malcolm Mitchell asks why policymakers target a low rate of inflation instead of deflation. This question is often asked by others, such as Austrian economists who argue that we should have a gold-backed currency that would be expected to lock countries in a mild deflation. I offer reasons why the current policy is reasonable. although I recognise that there is a cost to the erosion of the purchasing power of monetary tokens over time.

Waiting To Make Purchases Not That Big A Deal

Mitchell argues,

The standard logic is: When people anticipate the lower prices of deflation, they delay buying, waiting for those lower prices, and this delayed buying negatively impacts the economy.

What he characterises as the standard logic is widespread; I believe that it showed up in famous economics texts (I am unsure of the origin) and has since been repeated continuously.

However, it does not appear to hold up to inspection. Consumers have a lot of experience with the reality that electronics fall in price, yet people will buy products as soon as they are released. People need to buy food on a regular basis, even though they know that most items will eventually be placed on sale.

I would guess that people may delay purchases if there is an extraordinary price cut forecast to occur, and they will similarly ramp up purchases ahead of a coming price jump (such as hikes in consumption taxes). But there seems to be little reaction to the mild trends we seen in the price level in the developed economies in the past 20 years.

As a result, the answer is found in other areas.

Wage Stickiness

I would argue that the primary reason to avoid deflation is wage behaviour. People have an aversion to nominal wage cuts.  This was long argued to be true (such as by Keynes), and it shows up in the data. The chart above (taken from the NBER article Some Evidence on the Importance of Sticky Wages, by Alessandro Barattieri, Susanto Basu and Peter Gottschalk) shows the distribution of wage changes in the economy, The distribution is obviously skewed away from negative wage changes; if people were not bothered by their wages falling, we would expect the distribution of wage changes to have a symmetric distribution.

If wage changes continue to follow this form of distribution, it implies that average wages will continue to rise, even if the median annual wage inflation is near zero. There are not enough incidences of falling wages to counteract the affect of rising wages within the average.

In order to generate a deflation in wages, it would require serious intervention to overcome this resistance. Such policies would not popular amongst workers, and workers are a rather large voting block. It would also be necessary to force companies to lower prices in line with wage and productivity trends. Otherwise, companies could keep output prices steady, and pocket the extra profits. (This has been happening in the current cycle.)

Technically, it would be possible to have average wages that are stable over time, yet consumer prices experience a mild deflation. This is because of productivity - workers generally produce more over time as a result of increasing capital and more efficient. However, many products and services are near their effective productivity limit, and in some cases, productivity is falling. The aggregate price level might fall, but that would only be the result of improvements to productivity caused by things like the introduction of new products.

Wage Targeting?

Since productivity is not stable, it would difficult to set a target consumer price inflation rate that is negative yet keep wages from deflation. Instead, it might be possible to target wage stability, and let consumer prices be sorted out by the market; the assumption would be that they would fall if productivity is growing. (Note that an individual could have rising wages over a lifetime as a result of gained experience.)

It would be necessary to support such a policy with non-monetary policy changes, such as:

• government wage scales would have to be fixed;
• minimum wages and things like a Job Guarantee income would have to be fixed;
• transfers like old age pensions would have to be fixed (de-indexed from inflation);
• the brackets in progressive income taxes would have be non-indexed, to create a tax disincentive to wage creep.

(Attempted price level stabilisation via control of the Job Guarantee wage is an aspect of Modern Monetary Theory that is under-appreciated. Other aspects of government policy would need to made coherent with this objective, as I note above. If positive wage growth was desired, dollar amounts would have to be inflated by that growth rate. Indexation to consumer prices needs to be avoided, however, as that creates positive feedback between inflation and fiscal policy.)

I find it interesting that such a policy target is not considered. One explanation appears to be that there are considerable difficulties for measuring private sector wages; governments are large employers and can have relatively standardised wage scales.Another is that this is a legacy of the Gold Standard; economists seem to have the idea that the monetary price of goods needs to be stabilised stuck in their heads. Finally, it is clear that central banks want to appear neutral in the struggle between capital and labour; if they are seen as directly targeting workers, the central bankers would make themselves an obvious political target.

Consumer Price Index Bias

Although there is a large body of people who are convinced that inflation is really 6% higher than the published CPI inflation rate, most economists who have carefully studied price indices argue that the CPI overstates the cost of living. In the United States, the Boskin Commission studied this, and some changes to the CPI were made as a result to lower the bias. (This led to the rise of the term "Boskinization" by TIPS traders, which is not complimentary.)

I do not want to be dragged into the debate about the level of the bias, but one could argue that about half of the 2% level of inflation target (which is standard in most countries) represents this upward bias. Therefore, central banks are much closer to targeting 0% inflation than the current inflation targets suggest. (Over 50 years, a 2% inflation rate will raise the price level to 269 when starting from 100, whereas a 1% inflation rate would only raise the price level to 164.) 

Price level targeting would actually imply mild deflation in reality as a result of these biases.

Peer Pressure Amongst Central Banks

Once other central banks started to target an inflation rate of 2%, a country would stand out if it had a different target. If a country announced that it was targeting deflation, currency traders and monetary hoarders might decide that the currency "was as good as gold" and pile into it. This would create a structurally overvalued currency. Some countries (like Switzerland) have an industrial mix that can do well with an overvalued currency, but most countries would face difficulties. Commodity exporters as well as low end manufacturers suffer if their currency is overvalued.

Better Alternative To Liquidation

One of the problems that does not show up in economic models that only have a single composite consumer good is that relative prices can get out of alignment. Industries can end up with output prices and industry-specific wages that are too high for the demand for their goods. As a result of the difficulty with cutting wages, it is hard for the industry to cut prices so that it can raise demand for its goods. 

The classical remedy to that situation is liquidation - let the misaligned industries go bust. However, the Keynesian response is that this just creates unemployment and excess industrial capacity, and there is little tendency in the economy to return to full employment. 

Instead, if there is a general price inflation, industries can keep wages an output prices constant, and this results in prices falling relative to the economy-wide price level. This allows industries to slowly dig themselves out of a hole.

Deflation Is A Bad Sign Cyclically

In my previous comments, I covered the reasons why long-term deflation was viewed as a negative. However, as noted above, getting to deflation is an extremely costly exercise. ("You can't get there from here.") We do not see many examples of deflation in modern economies (if we strip out the temporary effect of oil price drops).

• Japan. People have been screaming about "deflation" in Japan for decades, but the price level is actually close to where it was in the early 1990s. 
• United States. The U.S. economy was devastated as a result of the financial crisis, yet core inflation remained stubbornly positive.  

The mixture of unexpected deflation and debt is a particularly toxic mix. Business profits are suppressed, and the terms of residential mortgages are based upon the implicit assumption that rising wages will make the household debt load easier to bear over time. If the deflation was expected, debt contracts would presumably be adjusted to take that into account (that is, lenders would have more strict terms for lending). This is why I did not mention debt as being a reason why we cannot have steady state deflation.  [Update: This paragraph was added in response to a heads up by Auburn Parks.]

As was argued in the recent post at Fictional Reserve Barking,

To summarize: Inflation displays inertia and peoples' expectations about the future cannot be dictated by the central bank alone. Basically, inflation is the result of the interplay of supply of demand for goods and services. When you have more demand than supply, prices and inflation accelerate; when you have more supply than demand, prices and inflation decelerate. It's that simple. That's the secret to understanding what creates inflation, barring the effect of any bottleneck issues.

Since it seems that central banks cannot just wave the 'expectations wand' to bring about deflation, we would need a mammoth shock to get there.  

Price Level Stability?

There is one advantage of price level stability - people will have nominal anchors for judging value. One of the interesting digressions in Piketty's 'Capital In The Twenty-First Century'  is his discussion of how price level stability allowed authors the ability to quickly describe the status of people based on their monetary incomes. This practice has dropped out of literature; wages and prices change so much over time that previous dollar figures become comical ("One million dollars!"). In order to describe someone's wealth authors now have to describe their (real) possessions.

If prices and wages are stable over a long period, it allows people to make judgements based on absolute prices, One could hope that this would make it easier to avoid things like house price bubbles, but that seems optimistic to me.

Moreover, price level stability is not the same thing as a "0% inflation target". If there are upward deviations to the price level - such as an inflation during a war - a 0% inflation target would attempt to keep the price level at the new, higher level. The most likely outcome would be an upward ratchet effect to the price level over time, which would still mean that prices would not be comparable over long time periods.

To achieve price level stability, it would be necessary to deflate the price level after the upward shock, as was common in post-war periods during the classical Gold Standard era. However, these deflations are costly, and are not politically sustainable now that the working classes have the right to vote. (There is considerable nostalgia for 19th century monetary policy amongst gold enthusiasts, but they do not have a credible explanation as to why workers would now act like their disenfranchised counterparts during that era.)

(c) Brian Romanchuk 2014