Karthik Narayan on Measuring the Effects of Unscheduled vs. Scheduled Monetary Policy Announcements

SHRUTI RAJAGOPALAN: Welcome to Ideas of India, a podcast where we examine academic ideas that can propel India forward. My name is Shruti Rajagopalan, and this is the 2025 job market series, where I speak with young scholars entering the academic job market about their latest research on India. 

Our fifth scholar in the series is Karthik Narayan, who is a doctoral candidate in Economics at Nuffield College and at the Department of Economics, University of Oxford.  His research focuses on monetary policy, macroeconomics and finance in developing countries.  

We spoke about his job market paper titled, Macroeconomic Effects of Scheduled and Unscheduled Monetary Policy Surprises.

We talked about how the Reserve Bank of India makes and announces its policies, its impact on interest rates, inflation expectations and output, measuring the impact of policy announcements, the Lucas Critique and much more.

For a full transcript of this conversation, including helpful links of all the references mentioned, click the link in the show notes or visit mercatus.org/podcasts.

Hi, Karthik, welcome to the show. It’s a pleasure to have you here.

KARTHIK NARAYAN: Well, thank you very much, Shruti. It’s a great pleasure to be on your podcast. I’ve been a great fan of it. I’m especially happy to be on it as a result.

RAJAGOPALAN: I was telling you just before we started recording that the job market series we do with young scholars is my favorite part of the year because I get to read so many new, cool papers. Yours is one of them. 

You look at something that we are all very obsessed with in both good times and bad times, in the sense of watching what the central bank and the central banker do, and how it’s going to impact the broader economy, right?

More specifically, you look at how financial markets react to a policy change from the central bank. The way you are looking at it is, you separate when it is expected—that is, when it’s a scheduled announcement—versus when it’s a sudden or unscheduled announcement, in terms of the central bank reacting to what’s happening more broadly in the economy or in global markets. What you show is that the scheduled rate decisions behave in line very much with what theory would predict. Tighter policy is going to depress stock prices. It’s going to slow down or decelerate growth. It reduces inflation over a longer period of time.

Unscheduled decisions, they don’t look quite the same as this exogenous policy shock that we think about, right? Your explanation is that they often happen alongside other news. Markets are reacting to two kinds of signals, both what’s happening in the rest of the world, and what the central bank is reacting to, and what the central bank also announces.

The broader takeaway, I guess, is more methodological. That is, if you really want to think about or measure the true power of monetary policy and monetary policy transmission through the rest of the economy, it’s better to look at scheduled moves rather than unscheduled moves. First up, did I get your very detailed paper largely correct? Second, can you tell us more about it now?

Measuring Causality Is Hard

NARAYAN: Well, firstly, thank you very much for that excellent summary of the paper. I would say you are getting exactly what I’m trying to convey as an implication and a takeaway for other economists working on measuring monetary policy, not just in India, but also in other emerging markets where measurement of the effects of monetary policy continues to remain a challenge even as people are trying to do a lot of work in trying to measure that more accurately.

I just wanted to take a step back and provide some broader context and motivation for what I’m trying to do in the paper. We see central banks all around us all the time reviewing and altering monetary policy. If you read the financial press, there is just a lot of commentary around this all of the time. Surprisingly, it’s extremely difficult to find rigorous estimates of the causal effects of the central bank’s policies, and particularly its monetary policy, and how that’s affecting the economy.

There have been a lot of papers that have been written, particularly in advanced economies, trying to make a lot of headway in this direction. If you just open up any issue of the American Economic Journal: Macroeconomics or the Journal of Monetary Economics, you’ll find that there continue to remain debates among economists about whether we are measuring the causal effects of monetary policy correctly, even in a data-rich environment like the US.

The problem is just much more in developing countries, where you don’t quite have the same granularity of data available at the same frequency or for the same time periods. The fundamental problem here is that a central bank is trying to set policy in as systematic a manner as possible, in response to what is happening in the economy. As a result of this, for an econometrician interested in cause and effect, it’s very difficult to learn anything meaningful from regressing central bank interest rates on outcomes of interest like output or inflation because the central bank is, at some level, reacting to both present, past, and expected future economic conditions.

If you’re going to do this regression, it’s very difficult to get a causal effect of the central bank’s monetary policies on the economy. Of course, this number is very important for a number of reasons, most particularly for policymakers themselves, because it’s important for them to know the magnitudes or the quantitative effects that their actions are having on the economy in order to arrive at a reasoned judgment about how they should be reacting to any incoming macroeconomic event.

Two Stories About Weak Transmissions

What I try and do in this paper is to make progress on measuring monetary policy and its effects on the macroeconomy in India. If you look at the past literature on this subject in India, the consensus is that it’s very difficult to find, empirically, effects that are economically meaningful and theoretically in the expected direction on aggregate demand in India. The academic consensus in the empirical literature is that it doesn’t have very significant effects on aggregate demand.

Now, there are two competing explanations for this. Some people argue that in India and in other developing countries, monetary policy transmission is weak for a number of structural reasons, including the fact that the fiscal authority isn’t necessarily behaving in a fiscally sustainable way in terms of the public debt. When interest rates rise and there is no debt sustainability practice in place, markets expect these higher interest rates to actually be serviced by higher inflation in the future. Even if central banks are raising interest rates, people don’t lower prices because they don’t trust the government’s commitment towards inflation-targeting or keeping inflation under control.

There’s the other view, which suggests that maybe we need to think more carefully, empirically, about the data that we are using to identify the causal effects of monetary policy before arriving at a decision in our minds about whether or not monetary policy is effective, in the sense that we think about it in a macroeconomic model, in that higher interest rates help to lower inflation and reduce the output gap and help stabilize the economy when it’s overheated. 

I provide evidence in favor of the second explanation. One thing I want to say here for context is that given the empirical challenge that I’ve laid out—which is the central banks act systematically, and so it’s very difficult for econometricians looking at the data to identify the causal effects of any change in interest rates on macroeconomic outcomes of interest—what the monetary economics literature has converged on as a way around this is to try and identify changes in interest rates that were unanticipated by the market, and to use these what are called “monetary policy surprises” to estimate the effects of central bank policies on the economy. 

RAJAGOPALAN: Karthik, what you’re saying is super important. The reason I think it’s important is, in between all of this market response and people responding, what we forget is in times of crisis—like, recently, we’ve had COVID or the current shock when it comes to tariffs and things like that. Before that, we’ve had the Global Financial Crisis.

What we expect is that the government has some policy levers or interventions that it can make to actually smooth out some of these disturbances or shocks that people are feeling, right? A big part of that is monetary policy, which means that the central bank is going to use the biggest tool it has, which is figuring out the interest rate and, therefore, the money supply. Through that, send a signal into the economy on how everything else will follow.

Because people are not chess pieces on a board, and they’re actually rational and they have expectations, they react to some of these signals. The way they’re going to react to a particular rate set, whether it’s an increase or a decrease in interest rates, is very hard to predict. The reason sometimes it’s hard to predict for the regular readers is sometimes that information is already included because of what’s going on in the market.

Sometimes it leaks. People are like, “Oh, MPC [Monetary Policy Committee] is going to meet tomorrow, and they’re definitely going to go to a rate cut. There are satta [betting] markets in India, which are betting on how big the rate cut will be,” and things like that.” Sometimes this information is already absorbed, and that’s why you don’t see a big reaction. Sometimes it’s not absorbed, and you see a big reaction, or sometimes you see a reaction over a period of time.

This becomes crucial for us to figure out how a central banker must operate in times of crisis and in good times. In some sense—sorry for the long intervention—but I feel like you’re almost underselling what you’re doing by making it a conversation between the literature of monetary economics. This really matters for every single person, every single checking and savings account, and what is going to be the cost of tomatoes tomorrow in the market, right?

NARAYAN: Absolutely. I agree with that point entirely, and particularly so, because if we don’t have accurate quantitative estimates of how the interest rates are affecting the economy, and the central bankers themselves don’t have a clear idea about this, then they may be acting too aggressively or not aggressively enough in response to incoming economic events. That has very direct, tangible economic implications for everybody who’s at the receiving end of that policy, which is, in the case that I’m studying, everybody living in India.

What Counts as a Policy Surprise?

RAJAGOPALAN: First up, how does one even go into thinking about this? What would be a monetary policy surprise versus what would be a scheduled announcement in this world of 24-hour news and tweets, and people monitoring cars going in and out of the RBI? What is a good way to even think about this as a setup?

NARAYAN: I’ll start with scheduled and unscheduled announcements first. What I mean by a scheduled announcement is any announcement that the RBI makes as a review of monetary policy that is part of its regular cycle. Today, we have the Monetary Policy Committee. The Monetary Policy Committee, we all know much in advance, meets six times a year. The dates are also advertised in advance, and a press conference is convened for this purpose.

Regardless of whether the RBI wants to make a change in monetary policy, the MPC will, in this scheduled interval, meet and review the monetary policy environment and decide if anything needs to be done. If it feels nothing needs to be done, people will just have a press conference saying, “We have decided to keep all rates unchanged,” which the RBI has done a lot of in the past few years.

Whereas an unscheduled announcement is one which is outside of this regular meeting cycle, where the RBI . . . Notable examples of this were during the COVID-19 pandemic. If you look at March 27, 2020, there was an emergency meeting of the Monetary Policy Committee. There was a sudden change in interest rates, which was completely unanticipated by the market, because the RBI felt the need to act to reduce the liquidity constraints that the economy was facing as a result of the lockdown.

RAJAGOPALAN: There are a number of different rates that we usually talk about, right? There isn’t one universal interest rate that we talk about. What is perhaps the best measure of looking at surprise announcements? I imagine it’s nothing very long-term. It’s got to be something quick, that’s easy to measure, that’s a quick reaction. How do you go about constructing the actual measure of how to think about the surprise?

OIS and MIBOR: Expectation Thermometers 

NARAYAN: How I construct a monetary policy surprise is using what is called an overnight indexed swap [OIS] contract. Now, I’ll explain what this is in more detail, but this is essentially an interest rate swap contract that is linked to what is called the Mumbai Interbank Overnight Rate. Now, the Mumbai Interbank Overnight Rate [MIBOR] is an interbank interest rate. It is the interest rate at which banks in India borrow from each other overnight.

This also is very closely related to what is called the weighted average call rate, which is another measure of the interbank interest rate that the RBI for construction maintains. Now, the weighted average call rate or the MIBOR rate serves as an operating target for RBI’s monetary policy. The RBI may have at its disposal several tools to set monetary policy.

Everyone would’ve heard of the repo rate, but there are many rates linked to that, like the marginal standing facility rate, the standing deposit facility rate. You have the cash reserve ratio, the statutory liquidity ratio. The RBI, in various instances, is going to be moving one or more of these interest rates to set monetary policy, but the end goal of all of these actions is to affect the MIBOR or the weighted average call rate. Whatever is happening to MIBOR serves as a summary statistic for the RBI’s monetary policy.

That is what I utilize to construct the monetary policy surprises. The way I do that more specifically is that, if you look at the overnight indexed swaps, what they essentially consist of are two legs. In an interest rate swap, you have two people deciding to swap interest rates. Now, one of those interest rates, because this is a MIBOR OIS, is going to be the MIBOR interest rate; that is market-determined on a daily basis. The other leg, which is often called the fixed leg of the OIS rate, is going to be determined by the two parties that are entering into this contract.

Now, typically, they enter into this contract for a fixed period of time. I’ll specifically be looking at the one-month MIBOR OIS, but there are MIBOR OIS rates going up to 12 months’ duration. When people are setting the fixed-leg interest rate for the OIS contract, what they’re essentially trying to do is forecast what the path of the MIBOR rate is going to be for the duration of the contract. In order to ensure there is no arbitrage opportunity available, they’re going to set an interest rate such that the expected cash flows from investing in MIBOR for that duration exactly equals the cash flows you get from this fixed interest rate on the OIS contract.

RAJAGOPALAN: Basically, you want to measure the expectations that the people who are closest to the financial system are setting or interpreting as a result of the policy announcement, right? This is not how I interpret how my mortgage or my car loan interest rate is going to be. That’s really far from the announcement. The OIS swap is about the closest. Because it’s a swap and their skin in the game, it’s probably the best way to measure the signal or the expectations. Is that a good way of thinking about it?

NARAYAN: Absolutely. I would also add to this and say that people who are usually trading these OIS contracts are banks and financial institutions who want some kind of interest rate protection against fluctuating interest rates. They take these overnight indexed swaps that give them the ability to swap their risky floating interest rate exposure for a fixed income return. Often, these participants are much closer and much more closely monitoring what the RBI is doing. You can think of these as the expectations of the RBI’s future path of monetary policy held by rather well-informed market participants who have a lot of skin in the game, as you said.

RAJAGOPALAN: Also, because it’s the large part of the banking system, it also affects what are going to be the downstream rates following that for regular folks and regular businesses, and so on.

NARAYAN: Absolutely, yes.

RAJAGOPALAN: Now that you can measure how to think about a surprise announcement through this very, very specific OIS, overnight indexed swap, how does one think about the impact through monetary policy? The two things that we typically care about as economists are output and inflation, right? Unemployment gets thrown into the mix from time to time. In India, a little bit less so. What are the standard and the nonstandard measures that you’re actually going to look at, let’s say, immediately and over a longer period of time to see if the announcement had the actual impact that it intended to have?

NARAYAN: In the monetary economics literature, economists look at a number of things, both at a very high frequency and at much lower frequencies, to understand how monetary policy transmits through the economy. At a very high frequency, even within the day, it’s possible to look at what the change in interest rates of the central bank is doing to bond yields of longer-maturity government bonds, what it’s doing to stock prices and other asset prices.

That gives you a good sense of the extent to which the interest rate changes are passing through prices and interest rates that matter for final decision-makers in the economy. A bit further down it’s possible to look at how these interest rate changes affect expectations of both professional forecasters as well as of households, and how they form expectations about how economic conditions are going to be in the future.

Now, expectations matter as an economic quantity in themselves. This is something central banks monitor very closely because people are forward-looking decision-makers. Their expectations of what future conditions are going to look like is going to have an impact on how they’re going to make their consumption and investment decisions. That’s going to translate into having an impact on GDP and on inflation.

Finally, of course, it’s possible to trace out what economists call an impulse response function to understand the dynamic response of inflation and of output quite directly in response to these monetary policy surprises. You could just take your measure of industrial production or some other measure of output. You could take your measure of inflation and regress that in period zero and then in period, say, two years or three years from now, directly on the monetary policy surprise to see to what extent it’s affecting the economy.

It’s possible to trace out an impulse response function, which shows how the effect of the interest rate change is affecting these variables over time. Typically, macroeconomists are very interested in these dynamic responses because that is what they’re going to be using to build their macroeconomic models, which central banks increasingly use both for their own internal assessments of how monetary policy affects the economy, but also for forecasting purposes.

RAJAGOPALAN: Now that you’ve explained that there are two parts to this—so one part is just asset prices, which you’re looking at, which is an immediate target. Also, there’s a lot of classification. The second part is the broader macroaggregates, output, inflation, and so on. If I come back to the asset prices, the way I think about the mechanism is that if the one-month OIS fixed rate is higher over, say, the two-day window, then markets will learn that policy is tighter than they had originally thought. If it’s lower, then they’ll think that it’s looser than they originally thought, right?

NARAYAN: That’s right.

Short term versus long term effects on asset prices

RAJAGOPALAN: Let’s say if there’s a one percentage point, like 100 basis points surprise tightening on a scheduled announcement date, what you find is that it increases the one-year bond, five-year bond, and 10-year bond at different levels. The yields for the one-year bond are roughly 28 basis points, for the five-year bond are roughly 19 basis points, and the 10-year bond are roughly 16 basis points. Overall, it’s going to lower the equity indices. You find that that’s about 3 percent. Other effects like exchange rates and other things are relatively small.

When I look at your paper, this to me looks like the immediate asset price reaction, both in the bond market and the equities market, right? Now, within the asset price world, how do you think about whether these are blips, which are also responding to other kinds of information simultaneously because the time period is so tiny, or whether there are longer-term effects for the same asset prices that you’re monitoring for the two-day window, and so on, so forth?

NARAYAN: That’s a great question. My response to that is that it’s possible to look at the longer-run responses of those same variables by doing impulse response functions, as I had mentioned earlier, where you take the same variable—say, take equity prices. You look at what equity prices are going to be in response to this interest rate change six months from now, nine months from now, 12 months from now, and you can plot those impulse response functions.

In particular, in my paper, what I show is that the equity price reactions to monetary policy surprises are starkly different on scheduled announcement dates and on unscheduled announcement dates. In particular, on scheduled announcement dates, you find equity prices declining on impact, which you see in the event studies on the same day. I also show later on in the paper that if you even do impulse response functions with these equity prices, you will find that these declines are rather persistent.

They last for around 18 months, 24 months after the change in interest rates. It is not a short-lived reaction. It’s something that has persistent effects on these asset prices. Whereas on unscheduled announcement dates, you actually find equity prices going up on impact. If you look again at the impulse response functions, they tend to remain so, for at least some period of time before returning back to their normal levels.

Now, one point I wanted to add here for context is that we are trying to measure monetary policy surprises using these interest rate swap contracts. The essential idea behind identifying these monetary policy surprises is that the change in the fixed rate that we are seeing around the monetary policy event is driven entirely by the monetary policy event and not from information coming from elsewhere.

Now, of course, one of the major points that I’m trying to make in the paper is that this is much less likely to be the case on unscheduled announcement dates than it is on scheduled announcement dates. Now, how does this connect to the asset price responses? Asset pricing theory tells us that any asset price is really determined by two quantities. One being the future cash flows that the asset is going to provide the holder, and the other is the discount rate at which they’re discounting this future cash flow.

Now, when interest rates rise, we know that credit conditions are tighter. It’s going to be harder to generate cash flows in the future. We would expect future cash flows to go down, and the discount rate is also rising. For both these reasons, we’d expect the asset price to fall. Now, to the extent that we don’t see it falling suggests that there is also other information coming out along with the monetary policy announcement on unscheduled announcement dates.

That’s what is perhaps driving the responses in a theoretically unexpected direction that we are seeing on unscheduled announcement dates. Just to give you an example, if we see on an unscheduled announcement date interest rates going up, say, 100 basis points, it could be that interest rates are going up because there is better-than-expected economic performance, and news about that is getting released at the same time.

Now, given the granularity of the data we’re having to work with in India and also in other developing countries, when you’re looking at the change in the fixed rate in this two-day window around the announcement, you’re picking up the effect that financial market participants are seeing as a surprise from the central bank’s own announcement, but you’re also picking up this new information that they’re getting at the same time about better-than-expected economic performance.

Now, better-than-expected economic performance, they’re more likely to revise upwards their expectations of future cash flows, their expectations of how the economy is going to perform. Due to all of these reasons, it’s hard firstly to get a clean monetary policy surprise on unscheduled announcement dates. Following on from that, it makes it hard to get a reasonable causal effect of the interest rate changes of the Reserve Bank of India on our outcomes of interest by using the surprises that we’re getting on unscheduled announcement dates.

RAJAGOPALAN: No, and I mean, also just from common sense, right? If it’s an unscheduled announcement date, it probably means something happened, right?

NARAYAN: Right.

RAJAGOPALAN: It could just be good news, and things are going well. The economy, more often than not, usually, they’re reacting to some adverse event, if not in India, globally, and everyone else is, too. I think just in terms of common sense, if there’s a meeting that’s called, that’s not one of your six usual meetings, something happened. And it’s no big secret because these are typically big important global events, or a new report has come out, or Standard & Poor’s has changed its rating on India.

There are so many things that could happen, but it’s a little bit naive for people to assume that central bankers will have all this information and the rest of the market participants will have none because, usually, it ends up being the other way around. There’s a very common-sensical explanation for why you do what you do, I guess.

Noise and Fiscal-Monetary Coordination

NARAYAN: Yes. Well, there’s just two points I want to make on this. The first is that while we see these problems with unscheduled announcement dates, unscheduled announcement dates are also very useful for an empirical macroeconomist because of the fact that you very rarely get very large changes in interest rates. Typically, on unscheduled announcement dates, you find very large changes in interest rates. 

For any econometrician working with macro time series, getting the kind of statistical power that you get on unscheduled announcement dates, it’s like a gold mine. The point I’m trying to make in the paper is that—

RAJAGOPALAN: It’s noisy. [chuckles]

NARAYAN: Yes, it’s noisy, and it comes at a cost of imprecise identification. You’re facing this tradeoff between power and identification when you’re deciding to use unscheduled announcement dates. Now, in many contexts, the number of unscheduled announcement dates that you would have in your time series would be very, very tiny. If you look at the US, for example, in the last 25 or 30 years, there have been far fewer unscheduled announcements than you have in India.

In India, there are just a very large number of unscheduled announcements over the 20-year period that I look at between end of 1999 and the beginning of 2020. Because of that, the problems of identification associated with using unscheduled announcements are just much greater in India than they are in other contexts and make it all the more essential to grapple with this tradeoff in that context.

RAJAGOPALAN: Yes, this is an important thing to flag. One of the reasons this is so recent in India is that monetary policy, as a serious tool for intervention, is a relatively recent post-liberalization phenomenon. You have to have really strong, robust markets for monetary policy to actually transmit anything meaningful, which really doesn’t quite work in a command-and-control economy.

They can set interest rates all they want, but you’re not going to get that dynamism that you can only have in a market economy, which, in India, starts really only in the ’90s. The fact that we’ve had only, say, 25 to 35 years of really good monetary policy instruments, and, in that, we’ve had only 20 years where we’re learning to do how this is done. Now, only a few years, less than a decade of the MPC, so that just gives you a sense of, we’re learning this as we go along, in some sense.

NARAYAN: Absolutely. The one other point I’d add is that in order for monetary policy to work well, you also need fiscal-monetary coordination. Often in India, in the pre-liberalization period, fiscal-monetary coordination often meant fiscal dominance of monetary policy, in that the government could issue debt. The RBI had an obligation to purchase that debt, which essentially meant the RBI had to print money to purchase whatever debt the government was issuing.

One of the implications of having that kind of system is that no participant in financial markets or, for that matter, any economic decision-maker is going to have any confidence that inflation is going to be anchored around a particular level, and that it will be low and stable regardless of any target that the government announces.

The really big policy reforms aimed at reducing and completely eliminating, for example, automatic monetization of deficits and providing central bank independence and a target for the central bank to target an inflation rate have all been steps towards establishing greater credibility in the eyes of market participants and economic decision-makers that the government is serious about keeping inflation stable and low. That has proven beneficial. If you look at the inflation record since the MPC and inflation targeting has come in, inflation broadly—if you leave out the COVID period—has been at levels that fall within the RBI’s mandated inflation target.

When the RBI Act was amended in 2016 as part of the Urjit Patel Committee recommendations, the Reserve Bank of India, which at that time was targeting what it called multiple indicators, in that it was trying to keep prices under control but also ensure a fast rate of economic growth, was suddenly told that “your only target is to ensure that consumer price inflation remains between 2 percent and 6 percent, with a point target at 4 percent. On average, you’re trying to maintain 4 percent, but you have leeway to go above or below, plus or minus 2 percent.”

Since that was introduced, the inflation performance—if you look at the CPI inflation rates since the introduction of inflation targeting, they’ve broadly been within that band. There are very few instances of the inflation rate going outside of that band because the RBI, according to law, is now mandated to write to the government to provide reasons in case it fails to meet that target.

That has, first, helped to anchor inflation expectations. People now take the mandate seriously. That is half the battle won because when firms setting prices are baking in the target rate of inflation, they’re going to raise prices by less. That’s half the battle won for the RBI already. Then it also makes all of its other communication and all of its other policy actions much more credible. In that sense, it’s been very successful as an experiment in the recent past.

Inflation Before and After the MPC

RAJAGOPALAN: For you, we have a secondary problem in the paper that you encounter in the sense that the MPC is a relatively recent thing. The RBI, as you said, got amended in 2016. We really have data only after that, so about nine years’ worth of data. The link between surprise announcements and asset prices is quite clear and quite tight, but the most important mandate, as you just discussed, is inflation.

What is a good way of thinking about these scheduled surprise policy announcements before the MPC and after the MPC was instituted, and how that has impacted inflation? Is there a major break that you see between the two periods? Is it more continuous in the way markets react, because the underlying mechanism is the same? What’s a good way to think about that?

NARAYAN: Well, I agree entirely with you in that the underlying mechanisms are the same. If you ask me about what is likely to have changed with the introduction of the MPC and the inflation targeting regime, it is really about inflation expectations. Prior to the introduction of an inflation target, the RBI was making no systematic effort to anchor inflation expectations— or for that matter, measure inflation expectations. It’s only in the recent past that there have been attempts to both measure household-level and firm-level expectations of inflation.

Secondly, to try and anchor them around the inflation target. Now, that is very helpful from the standpoint of monetary policy transmission to inflation, because if people are anchored around the inflation target, inflation is just going to be less sticky and fall much more quickly from whatever level it is at to the target level, much faster than it does in a world where you don’t have any anchoring of inflation expectations.

I do have some evidence in my paper in the appendix on how the response of CPI inflation has changed rather dramatically with the introduction of the Monetary Policy Committee and inflation-targeting regimes, with the prices reacting much more immediately and swiftly in response to interest rate changes. It does have an effect, and I have some evidence to show that it does. Of course, using more recent data would allow us to establish this more clearly.

RAJAGOPALAN: You could do a part two of this as we go along in time. I have a question which is a little bit about how the sausage is made. If you are on all these policy economists and geeky economists WhatsApp groups like I am, then you know that before there is an MPC scheduled meeting, people are like, “Hey, what do you think is going to happen? All the economists are chiming in on what they think is going to happen. Twitter is going crazy.”

The central banks all over the world are not exactly known for keeping secrets. I won’t exactly call them leaks. It’s not actual papers getting leaked, but central bankers are talking to people in the markets all the time to get a sense or a pulse of what’s going on, right? How much of all of this is like, in one sense, almost endogenous? Central bankers know what you are saying, which is scheduled announcements are understood quite differently from unscheduled announcements.

They already have this information. They know how the markets are going to react. They have a rough sense of the asset price reaction versus the inflation reaction. Now they themselves adjust what they’re going to say, and then the market players know that they’re already adjusting what they’re going to say based on the reaction, so they react and adjust there. This is almost like a crazy, dynamic back-and-forth that keeps going on, right?

NARAYAN: Absolutely.

RAJAGOPALAN: What is a good way to think about that, given everything else that you’ve told us?

NARAYAN: One thing I’d say in terms of measuring the monetary policy surprise itself, the way the method I’m using gets around this problem of there potentially being side channels of communication between the central bank and financial markets is simply the fact that, to the extent that such communication exists and to the extent that there is valuable information about monetary policy in that communication, it’s going to reflect in the fixed rate getting appropriately adjusted. Market prices are going to reflect not just publicly available information, but also potentially privately available information that they have from the central bank.

RAJAGOPALAN: That’s why you use the OIS, because the swap will tell you everything you need to know in some sense.

The Lucas Critique

NARAYAN: Exactly, so that’s one thing. The other point I want to make here is that the responses that I am showing in the paper are responses to these monetary policy surprises. For listeners who’ve heard of the Lucas critique, they will know where I’m going—is that it’s very difficult to extrapolate from the responses that we are seeing to the monetary policy surprises to saying how systematic monetary policy will go on to affect the economy.

What do I mean by the Lucas critique? The Lucas critique, particularly in the context of what I’m saying, is essentially the idea that you simply cannot extrapolate to look at the effects of a particular policy in a counterfactual sense, if it’s different from the policy that you’re implementing.

Just to give you an example, if I’m providing you with estimates for how the economy responds to a 100-basis-point change in monetary policy under an inflation-targeting regime and, tomorrow, if I were to make an assessment of how the economy is going to react to a 100-basis-point change in monetary policy, when I have a regime in which there’s automatic monetization of government deficits, I simply cannot use the same number because I need to take into account the fact that everybody in the market is going to update their expectations about how inflation is going to evolve in the future because there has been this change in fiscal-monetary coordination.

Essentially, therefore, the idea that you cannot really say anything about counterfactuals without first building in a model that has agents, who are also taking into account the changing rules of the game in that counterfactual world that you’re looking at. That is typically what macroeconomists do with these estimates that I’m generating, in that they will try and write a macroeconomic model, what’s often called a dynamic stochastic general equilibrium [DSGE] model, where you would have households, firms—you’d have the central bank also as an agent as a part of that model.

They will then specify all of these agents with expectations about how the economy will evolve. These expectations will, of course, be the actual expectations that are driving random processes inside the DSGE model. Then they will try and use that model to match the estimates that we’re seeing in the data, and then use that calibrated version of the DSGE model to make other counterfactual assessments about how policy will evolve under different circumstances.

So connecting that back to my paper, what I essentially wanted to say was that these are responses to monetary policy surprises. In order for this to be useful for a policy audience, the direct connection comes from integrating this with a DSGE model that, say, the RBI itself has been using in the recent past. Updating the parameters that they’re using to calibrate that model and get a better assessment of how their own policies are operating in the economy, and also get better forecasts for how the economy is going to evolve in response to the various actions that they’re taking. That’s how it connects, and that’s how modern macroeconomists make use of these causally identified moments in the data.

Practical Implications

RAJAGOPALAN: If I had to ask you, “What is the policy implication of all of this?” Very directly, when I think about it, your paper is really telling us about measurement and how we think about measuring monetary transmission. Now, let’s say I’m a central banker. Now, what is the message that I should take away? Is it that I shouldn’t have unscheduled policy announcements?

That seems crazy because the whole point of having a monetary policy lever is that you can actually use it when things are really good or really bad. That is when you need to temper things in an exuberant market, or you need to actually propel things because of an adverse global event. What is it we can really take away from your paper in a more pragmatic playbook/manual sense?

NARAYAN: One thing I’d say here, to the RBI’s credit, [that] it has already internalized in its decision-making procedures, is the fact that there is great value to reviewing monetary policy very frequently. If you look at the RBI’s own 20-year history on this, in the early 2000s, the RBI was conducting monetary policy scheduled reviews twice a year before the two planting cycles that were there in India at that time and continue to be so. Back then, it was a much more agrarian economy than it is today.

The agricultural cycle formed the basis for when the RBI was reviewing policy. As India’s economy has industrialized, opened up, and become much more globally integrated, global events have much more immediate implications for India’s economy. As price controls have gone, shocks are going to start reflecting on people’s wallets much more quickly. All of this means that it’s much more essential to act quickly and, therefore, to be on the lookout for events that threaten macroeconomic and financial stability.

The best way to do that is to have regular scheduled reviews of monetary policy. At least since 2016, the RBI has been doing a review every other month, much in line with what is happening in the rest of the world. For the econometrician, that’s very helpful because, firstly, it eliminates a lot of the unscheduled announcements, because when you have very frequent scheduled meetings, there is less of a need to act outside of that schedule.

It’s also beneficial from a macro and financial stability standpoint because the central bank is communicating at regular intervals and is also putting together the institutional resources to review monetary policy at regular intervals, both of which are very helpful. Even if the central bank were acting outside of schedule to deal with immediate developments, it’s always much better to make these decisions once you had a team of economists put together an economic policy report to analyze the economy and have deliberations before making that decision, rather than acting in haste when something erupts.

For both those reasons, I’d say the major policy lesson, which at some level has already been learned in many countries, is that you should review policy very frequently. An important ingredient to that is having a statistical system in place that gives you sufficiently high-frequency information about the economy. That is a prerequisite for having these regular scheduled meetings at, say, a monthly or a bi-monthly frequency, is that you need to have new information to process at each of these meetings. Setting up that statistical infrastructure is also an equally important component of this.

RAJAGOPALAN: As more Indians get plugged into the banking system, as more transactions become more and more formal as opposed to what was the underground economy in India during command-and-control systems, the better data we’re going to have. Then these data can also be sliced and diced in more granular ways to actually tell us what is it that a central banker or anyone else is measuring. This was really fun to read. “Fun” is a weird word to use for this paper. I actually learned a lot because it’s such a detailed model, and I don’t encounter this stuff frequently. It was super fun to actually read this paper because we’re living through this in India as we go along, right?

NARAYAN: Absolutely. I think that getting the macro and financial stability right is very essential for any growing economy because no investor is going to want to come in and invest in an economy where the underlying institutional features—the institutional plumbing that’s keeping both the economy and the financial system going—if they aren’t being handled well, people will think twice before investing.

RAJAGOPALAN: Absolutely.

NARAYAN: This is the most robust foundation that’s necessary in order to sustain economic growth at, say, double-digit levels for a sustained period of time.

Other Research Interests

RAJAGOPALAN: Which is the hope. One last question before I let you go: What else are you working on? Can you tell us something about some of your other papers or your other research interests?

NARAYAN: I am very interested in monetary policy and, in particular, how it works in India. In one of my other papers—and this is coauthored with Vimal Balasubramaniam at Queen Mary University of London—I look precisely at the point you made earlier about what role growing financial inclusion has in sustaining monetary policy transmission.

What happens to the transmission of monetary policy to household income and household consumption as these households get plugged into the banking system? As you get plugged into the banking system, you become much more sensitive to interest rates because your fixed deposits are going to change in terms of the interest rates, and those deposits are going to change.

When central banks change their policy rates, if you’ve taken a mortgage or if you’ve taken a loan, the EMIs [equated monthly installments] you’re going to pay on that are going to react much more quickly. They’re going to react in response to these interest rate changes. Therefore, you’re going to be affected much more immediately in response to any interest rate change. In that paper, we try to understand this in more detail by putting together these monetary policy surprises with administrative data on financial access in India and household survey data on consumption and income. That’s one other paper I have related to this. That is a work in progress.

There’s another piece of work I have—again, a work in progress, coauthored with Aeimit Lakdawala and Rajeswari Sengupta—which looks at central bank speeches and the information that they may contain, and how they transmit through the economy as well. These are a couple of other projects related to this that I’m working on at the moment.

RAJAGOPALAN: That sounds fascinating, the central bank speeches especially, because the entire financial press will show up. We’re all looking at it, reading every word and every eyebrow movement with hawkeyed attention. This sounds fascinating. Thank you so much for doing this, Karthik. This was such a pleasure.

NARAYAN: It was a great pleasure for me to be on the show as well. Thank you very much for taking the time to read through my work, as well as to engage with it so thoughtfully. Thank you.

RAJAGOPALAN: We hope you come back.

NARAYAN: Well, I hope to be back also. Thank you very much.