Saturday, March 07, 2009

Watch the policy rate in real terms, but worry about how little it matters

I have an article in Financial Express titled Watch the policy rate in real terms, but worry about how little it matters where I discuss monetary policy in India's slowdown.

A key flaw that is being made in the public discussion of inflation is the use of year-on-year measurement of inflation. While the yoy WPI inflation stands at 5.25% in January, the WPI has dropped from a level of 241.5 in September 2008 to 229.6 in January 2009. What we have today is a period of deflation but the use of yoy inflation measures - which represent a moving average of the latest 12 monthly shocks - yields a misleading input into economic analysis.

In similar fashion, CPI-IW inflation shows 10.45% on a yoy basis in January, but the level of the CPI IW has been flat from October 2008 to January 2009. To use yoy inflation as a measure is to get a very wrong picture of the inflationary pressures in the economy.

On the subject of better early warnings of inflation, and how this sheds light on monetary policy in the past, see this paper.

7 comments:

  1. hi sir,

    In your article in the Financial Express, you've mentioned about feeble monetary policy transmission in India, which I presume is because you think that interest rate changes (small or big) are not really having the kind of impact on the overall economy as they do in the West, more specifically in the US. Is this presumption correct?

    if the presumption is correct then, how does a system that is smooth in terms of its monetary policy transmission (i.e. US) affect a household that has ZERO or very little debt/liabilities (which is what a large part of Indian middle class is)?

    The reason I ask this is because today changes in interest rates are having a significant impact on a certain segment of the population - i.e. the ones with home loans, car loans, credit card loans...etc.

    When consumer debt barely had a presence, say about 10 years back, there was little or no impact on the household cash flows. Now, that consumers have started borrowing, even small changes in interest rates are affecting household cash flows, which in turn means that the monetary policy transmission is working atleast in case of those are in debt.

    Does this mean that monetary policy transmission works effectively wherever there is debt in play, but may not or need not work directly when there is little or no debt, all this with or without the BCD nexus.

    ReplyDelete
  2. The monetary policy transmission is effective when (in the bottom line) a relatively small cut in interest rates yields a significant impact on aggregate demand (and vice versa). Suppose we had a proper business cycle coincident indicator in India. Then the question would be: When RBI cuts rates by 25 bps, does this yield a significant impact on this indicator over a few quarters?

    The channels are many. Household credit is one of them but it is only one of them. You are right in thinking that, other things being equal, lower indebtedness of firms and households reduces the effectiveness of the monetary policy transmission.

    To some extent, the leverage of firms and households is itself shaped by the macroeconomic environment. When there is macroeconomic stability, people are willing to take more debt. When there is a more difficult macroeconomic environment, with weak fiscal/financial/monetary institutions, individuals have less confidence and take on lower debt. So we shouldn't think of the leverage found in the economy as exogenous to the institutional framework.

    ReplyDelete
  3. A lot of questions:

    "Seasonal adjustment for inflation"

    http://rbidocs.rbi.org.in/rdocs/Bulletin/PDFs/79795.pdf

    Why would RBI estimate these seasonal factors and not use them. Not clear to me. Are people in RBI that incompetent?

    "From August 2008 to January 2009, the WPI shows annualised inflation of -15%. From October 2008 till January 2009, the CPI-IW shows inflation of 0. This gives a policy rate -- in real terms -- of roughly 19% going by WPI and roughly 4% going by CPI. In other words, when the downturn came, RBI sharply tightened monetary policy - raising the real rate from 0 to a value between 4% and 19% depending on what inflation measure is to be trusted."

    1. Real rate is not an ex-post concept. It is an ex-ante concept. Unless you can say that expectations were realized over this period, your statement is conceptually improper. I see no reason why expectations will be realized in this market when we know there is systematic bias in pretty much all markets.

    2. Central banks look to pin down long-term expectations of inflation. No one bases policy on what the headline index is doing. Most focus on the core.

    Just because prices are falling does not mean they will continue to do so. Just because oil prices have fallen, does it mean they will continue to fall?

    Even if we assume that people are forming expectations about the future by observing inflation readings, why would people base price expectations for 5 to 10-years on headline for 4-12 weeks. Central Banks aim to pin down a suitable moving average (and of course to avoid excessive output vol.).


    "But it is also true that monetary policy is all about forecasting inflation and forecasting growth, and using these values to set the short-term rate. And macroeconomic forecasters will always be judged by their average forecast performance. For RBI to become a well respected central bank, it needs to embark on the process of building reputational capital by coming out right on these calls."

    So which central bank is well respected according to you based on "forecast performance"? Do you know any research that shows that the track record of macro-econometric forecasting is great anywhere in the world? I have not, but I doubt it.

    "Reasonable values for forecasted inflation now range from -5% to 0%."

    There certainly is no reliable survey measures of inflation expectations available. What model for inflation expectations have you used?

    "And monetary policy must signify something. It must reach out and influence asset prices all across the economy, through a well functioning monetary policy transmission."

    Monetary transmission is dis-functional in most parts of the world today. In India, there certainly are additional reasons.

    "While RBI has cut the short rate, this has not influenced much in the economy."

    Again true for most of the world today. You have to consider the overall picture. India does not exist in a vacuum. In the current environment, even if policy transmission was great and rates were cut even more, I doubt there would have been any traction.

    "Interest rates for corporate bonds have risen sharply in real terms."

    Again, not specific to India.

    "To achieve a strong and effective RBI that matters, far-reaching changes in Indian finance are required. A well functioning Bond-Currency-Derivatives Nexus, and banking reforms, are required. A well functioning Bond-Currency-Derivatives Nexus, and banking reforms, are required."

    We have had this debate earlier. You still have to explain to me as to why monetary transmission would not work in a world where there is only one currency (so currency cut out of your nexus) and no derivatives (derivatives out of your nexus).

    Most derivatives are priced to permit no arb with cash. Cash has to get it right first. Just because you start trading derivatives does not mean that somehow magically cash will be priced correctly.

    U.S. 2-year swap spreads went hay-wire because cash inter-bank market went for a toss. Thus the solution lay in getting cash markets right and once the underlying normalized so did the derivative.

    In a lot of countries, a term interbank cash market fails to develop for various reasons. The market is often so one sided that it has a huge liquidity premium which swamps whatever expectations of the path of short rates it is pricing. And that is the reason that only OIS is quoted.

    You can't start quoting 5-year swaps against 3-month interbank term cash rate and expect that suddenly the underlying market would develop because of that.

    ReplyDelete
  4. Don't bother, Econologic. Mr. Shah is not going to respond to arguments he can't win, but then why should we hold him to a standard of scientific rigour?

    His logic here has been repeated many times: the transmission mechanism doesn't work as we don't have a BCD nexus, but the RBI still gets it wrong. Doesn't this beg the question? If whatever the RBI does cannot be transmitted to prices, how can it get anything right or wrong? It simply doesn't matter. Now, if it can affect things, some transmission must be happening.

    But again, I don't expect such glaring gaps in logic to be addressed, as consistency is a scientific virtue, not a political one.

    ReplyDelete
  5. Anonymous, this blog is only for quiet and rational discourse. Ad
    hominem attacks are strictly prohibited. The only reason I am not
    deleting your comment is because the attack is directed at me. :-)

    Yes, there is a problem in reconciling the two views: RBI doing the
    wrong thing and the monetary policy transmission being feeble. Do I
    contradict myself? Very well, I contradict myself. I am large, I
    contain multitudes.

    The monetary policy transmission is feeble but it isn't
    non-existent. So what monetary policy does matters. I would like for
    monetary policy to do the right thing. I'm sure you would, too. But
    when we get into this somewhat angels-on-pinheads discussion, we
    should remember that the stakes are rather low, for the monetary
    policy transmission is quite feeble.

    I thought I was trying to get that point across in the FE article. The
    very title says this: "Watch the short rate in real terms, but worry
    about how little it matters". I am constantly trying to keep both
    points in perspective.

    Econlogic:

    >A lot of questions:
    >
    >"Seasonal adjustment for inflation"
    >
    >http://rbidocs.rbi.org.in/rdocs/Bulletin/PDFs/79795.pdf
    >
    >Why would RBI estimate these seasonal factors and not use them. Not
    >clear to me. Are people in RBI that incompetent?

    I am aware of work at RBI on this. In this paper, we
    mention an RBI committee report headed by B. Singh.

    Why would RBI do work on seasonal adjustment and then not use it?
    Search me! Read a typical RBI document, and you see the ubiquitous use
    of year-on-year change of IIP or WPI or whatever. We just don't see
    them talking about "SAAR" (seasonally adjusted annualised rate) or
    "POP SA" (point-on-point seasonally adjusted) values.

    The above paper has some very telling tables where it looks like RBI's
    actions were based on YOY values and not what was actually going on in
    the seasonally adjusted series. One can, of course, never tell for
    sure about these things, but do look at the tables. They are quite
    remarkable.

    >"From August 2008 to January 2009, the WPI shows annualised inflation
    >of -15%. From October 2008 till January 2009, the CPI-IW shows
    >inflation of 0. This gives a policy rate -- in real terms -- of roughly
    >19% going by WPI and roughly 4% going by CPI. In other words, when the
    >downturn came, RBI sharply tightened monetary policy - raising the real
    >rate from 0 to a value between 4% and 19% depending on what inflation
    >measure is to be trusted."
    >
    >1. Real rate is not an ex-post concept. It is an ex-ante concept.
    >Unless you can say that expectations were realized over this period,
    >your statement is conceptually improper. I see no reason why
    >expectations will be realized in this market when we know there is
    >systematic bias in pretty much all markets.

    Yes, the real rate is an ex-ante and not an ex-post concept that the
    real rate is computed based on _expectations_ at a point in
    time. Expectations do not have to be fulfilled for them to be correct.

    E.g. if the broad market today thinks inflation is going to be 10%,
    then a short rate of 12% yields a real rate of 2%. If, later on, there
    are inflation surprises and inflation turns out to actually be -10%,
    that is irrelevant. The measurement of the real rate at a point in
    time requires a sense of what are inflationary expectations at that
    point in time.

    Ideally, this would be done using the gap between inflation indexed
    bonds and nominal bonds. But given the absence of a sensible BCD Nexus
    in India, that can't be done. So one is down to heuristics.

    Yes, I am handwaving, but I think you will agree that as of
    July/August 2008, most of us were thinking in terms of roughly 8%
    inflation. And the short rate was roughly 8%. So that gave a real rate
    of roughly 0.

    Where were we after Lehman died? Different people had different
    views. I think quite a few people thought this was going to break the
    back of inflation and significant deflation was on the anvil. We
    certainly saw that happen in many other countries worldwide (inflation
    expectations shifting dramatically). I do not think it is too much of
    a stretch to say that in India also, the sensible people looking at
    the series saw something big was afoot.

    Today, depending on whether you do ARMA models of CPI or WPI, you get
    forecasts ranging from 0 to -5% (SAAR). This gives you a real rate of
    3.5% to 8.5% depending on what you want to believe. So monetary policy
    is more tight today as compared with July/August 2008.

    >2. Central banks look to pin down long-term expectations of inflation.
    >No one bases policy on what the headline index is doing. Most focus on
    >the core.

    Not true. The typical inflation targeting central bank is told to deliver
    on what consumers transparently understand: headline CPI.

    >Just because prices are falling does not mean they will continue to do
    >so. Just because oil prices have fallen, does it mean they will
    >continue to fall?

    There is a lot of AR dynamics in prices. So yes, time series
    extrapolation is modestly useful.

    >Even if we assume that people are forming expectations about the future
    >by observing inflation readings, why would people base price
    >expectations for 5 to 10-years on headline for 4-12 weeks. Central
    >Banks aim to pin down a suitable moving average (and of course to avoid
    >excessive output vol.).

    I don't think I was discussing 5-10 years. I was discussing the short
    rate expressed in real terms, which is more like horizons of 5-10 weeks.

    >"But it is also true that monetary policy is all about forecasting
    >inflation and forecasting growth, and using these values to set the
    >short-term rate. And macroeconomic forecasters will always be judged by
    >their average forecast performance. For RBI to become a well respected
    >central bank, it needs to embark on the process of building
    >reputational capital by coming out right on these calls."
    >
    >So which central bank is well respected according to you based
    >on "forecast performance"? Do you know any research that shows that the
    >track record of macro-econometric forecasting is great anywhere in the
    >world? I have not, but I doubt it.

    I think the Bank of England has done a pretty good job of forecasting
    inflation.

    There are some papers out there looking at the forecast accuracy of
    inflation targeting central banks, and the results are impressive.

    >"Reasonable values for forecasted inflation now range from -5% to 0%."
    >
    >There certainly is no reliable survey measures of inflation
    >expectations available. What model for inflation expectations have you
    >used?

    Answered up there. Handwaving plus time-series analysis of CPI and
    WPI. Yes, this is not survey-based expectations, but I'm an optimist
    and think that market participants generally know a lot, and generally
    know more than me. So if I know something, it's surely in the market's
    expectations.

    >"And monetary policy must signify something. It must reach out and
    >influence asset prices all across the economy, through a well
    >functioning monetary policy transmission."
    >
    >Monetary transmission is dis-functional in most parts of the world
    >today. In India, there certainly are additional reasons.

    Agreed, transmission is broken everywhere :-) But in India it has
    always been broken, while elsewhere it used to work and went
    dysfunctional in the last six months.

    >"While RBI has cut the short rate, this has not influenced much in the
    >economy."
    >
    >Again true for most of the world today. You have to consider the
    >overall picture. India does not exist in a vacuum. In the current
    >environment, even if policy transmission was great and rates were cut
    >even more, I doubt there would have been any traction.

    The difference is that a dysfunctional transmission is the norm in
    India and the unusual circumstance elsewhere.

    >"Interest rates for corporate bonds have risen sharply in real terms."
    >
    >Again, not specific to India.

    Again, as above.

    >"To achieve a strong and effective RBI that matters, far-reaching
    >changes in Indian finance are required. A well functioning
    >Bond-Currency-Derivatives Nexus, and banking reforms, are required. A
    >well functioning Bond-Currency-Derivatives Nexus, and banking reforms,
    >are required."
    >
    >We have had this debate earlier. You still have to explain to me as to
    >why monetary transmission would not work in a world where there is only
    >one currency (so currency cut out of your nexus) and no derivatives
    >(derivatives out of your nexus).

    I apologise if I have been obscure.

    For the monetary policy transmission to work, when there is a change
    in the short rate, arbitrageurs must transmit it all across the curve
    (this requires a monetary policy rule). And it must transmit into all
    credit curves. Bank lending rates must move similarly to the corporate
    bond market. Stock prices must go up.

    All this requires an intricate system of arbitrageurs carrying small
    changes in the short rate all across the system.

    Derivatives and currencies are integral to this.

    >Most derivatives are priced to permit no arb with cash. Cash has to get
    >it right first. Just because you start trading derivatives does not
    >mean that somehow magically cash will be priced correctly.

    Derivatives on the yield curve are essential to getting an arbitrage
    free yield curve. It can be done without it, but that is not how it's
    done.

    ReplyDelete
  6. 1. My Comment: Central banks look to pin down long-term expectations of inflation.No one bases policy on what the headline index is doing. Most focus onthe core.

    Your Response: "Not true. The typical inflation targeting central bank is told to deliver on what consumers transparently understand: headline CPI."

    Yes, I agree, the aim is to deliver on headline CPI over a reasonable period of time.

    But "Basing Policy on" and "Delivering on a Target" are two different things.

    You are suggesting that short-rates should be cut or hiked based on the likely course of sa pop prices over the next 4-5 months.

    That is wrong. No central bank operates like that. Policy should not respond to transient shocks. Therefore, inflation prognosis is based on the core. No central bank rushes to cut or hike rates because the headline number is rising or falling over a short period.

    What matters is where long-term inflation expectations stand. Monetary policy is mainly about stabilizing long-term inflation expectations around the target. Short-term readings matter only to the extent they affect people's expectation formation process.

    2. "Yes, the real rate is an ex-ante and not an ex-post concept that the real rate is computed based on _expectations_ at a point in time. Expectations do not have to be fulfilled for them to be correct."

    Okay let me be slightly more precise. Expectations should be unbiased for your analysis to hold. There should be no systematic mistakes in expectation formation.

    P(t+1) = E(t)P(t+1) + mu

    mu~(0, sigma^2)

    Where, E(t) is the expectation formed at time (t) for price at time (t+1)

    If this were true, then there should be no forward bias. But we know that this is not the case in many markets.

    Why should it hold in your case?

    Depending on which school of thought you belong to, you would call it a risk premium or a behavioural bias.


    I have more to say and will post later.

    ReplyDelete
  7. Mr. Shah,

    Can you please explain why BCD nexus is necessary for effective monetary policy transmission.
    As I understand, the problem with India is that whenever RBI changes rates , our banks don't change lending rates and so the rates in the economy don't change. That probably is due to fact that there is no bond market in India so there is no alternative to raising loans from banks and thus its not the market forces determining rates there. But how do Currency and derivatives come into picture. Please dont point me to Mistry's or Rajan's report. I have read them and they just talk about what BCD nexus is. The question I want answered is how is this concept related to monetary policy transmission and also why monetary policy transmission in India is not as good as in US...whats the issue with India?

    Thanks
    Ajay

    ReplyDelete

Please note: Comments are moderated; I will delete comments that misbehave. The rules are as follows. Only civilised conversation is permitted on this blog. Criticising me is perfectly okay; uncivilised language is not. I delete any comment which is spam, has personal attacks against anyone, or uses foul language.