The title of my presentation is not very catchy. Catchy titles should fit on one line. I think my research papers tend to have un-catchy titles, which is probably why I am in central banking and not politics. But I think it conveys what I want to say. Has something changed with negative euro area rates? Are the spillovers that we see in countries around the euro area the same as before or are they different? I’ll try to highlight what I consider to be normal spillovers. I will then discuss some channels through which one may experience deviations from normal spillovers with negative rates based on some of my research. In my remarks, I am going to try to address three points, that were raised by Governor Sejko in his opening address.
The first is the effect of the change in relative yields across countries due to negative euro area interest rates. The second point I’d like to discuss is how low inflation in the euro area may be spilling over to other countries in the region. The third point is the effect of negative interest rates on capital flows, especially given the high concentration of euro area banks in the region.
With that, I am going to start with this slide (Figure 1), which shows the ECB policy actions since 2010. If we want to understand what has changed with negative euro area rates, from a theoretical perspective, we need to argue why the transmission of spillovers changes with negative interest rates. A potential mechanism runs through the effective lower bound on nominal interest rates. There is also an empirical challenge. In the figure above, the blue line shows the ECB deposit rate, the red line shows the size of the ECB balance sheet, and the two black vertical lines pick out the two points when ECB interest rates were decreased first into negative territory and then cut again. What you see is that it is pretty hard to empirically isolate the pure effect of negative rates from ECB balance sheet expansion. So, I am not going to do that here. However, I will try to illustrate how spillovers may change with negative interest rates.
I am going to base part of my talk on a paper I have written with Mick Devereux from the University of British Colombia and my colleague Giovanni Lombardo, mentioned in Figure 2. As you see, another paper with a not very catchy title, but basically what we do in this paper is to first establish some empirical facts about monetary policy spillovers. In this case, we looked at US monetary policy spillovers to emerging economies, principally in Asia or Latin America. We then explain these facts by adding certain types of financial frictions to a standard two-country Dynamic Stochastic General Equilibrium (DSGE) model. In our model there are two frictions that seem to be important in explaining how US interest rate changes affect emerging economies.
First, there is something that we call a double banking friction. That is, the banks in the center country, for example, the euro area, face a collateral constraint. When interest rates are cut, this raises the net worth of the banks, which in turn raises not only their domestic lending, but also cross-border lending to banks in emerging economies for example, Southeastern Europe. The second aspect of this spillover is that many emerging economies have foreign currency debt, and so movements in the exchange rate caused by interest rate changes affects the relative value of assets and liabilities, which amplifies the cross-border lending spillover.
In our model, we show that with these frictions there is little advantage from following an inflation targeting rule over a fixed exchange rate rule. So, it is interesting given the talks we heard yesterday that although countries in the region have quite different monetary arrangements: some have floating exchange rates, such as Serbia or Albania, others have currency boards – Bosnia – and some are completely euroized, the outcomes do not seem so different. This may be partly due to the effect of these financial frictions, so that there is little to be gained from one monetary arrangement over the other, at least in terms of short-term macroeconomic stabilization and spillovers.
Let me turn to characterizing normal spillovers from monetary policy. In Figure 2 above, we estimate the effect of a cut in US policy rates using narrative monetary policy shocks from Romer and Romer before interest rates hit the effective lower bound. Overall, a reduction in US policy rates generates an expansion in emerging economies. But where is this coming from?
First, the exchange rate in the emerging economy appreciates. Associated with the appreciation is a fall in the inflation rate of the emerging economy, potentially driven by the exchange rate appreciation. There is also an emerging economy policy response. When the US cuts the policy rate, other countries also cut rates and so they seem to follow the policy of the center country. That is the first block on the right there, where following the center country policy seems to be the standard response.
The second part covers the effect on gross capital flows. On the one hand, when interest rates are cut in the center, there is an increase in capital inflows into emerging economies. But, there is also an increase in capital outflows from emerging economies to the center. This is what we call a retrenchment effect. The increase in capital flows in both directions suggests that the banking or the financial sector plays an important role in transmitting monetary policy spillovers not only toward emerging economies but also to the center country. I am going to talk about how spillovers may change with negative interest rates in these two blocks.
The first thing that changes with negative rates is that a country may not be able to follow the leader anymore because it hits the effective lower bound. The second point, which seems to be different from a normal monetary policy cut, is that any compression in the banking sector profitability following negative rates may weaken cross-border spillovers through the banking sector.
Has anything really changed with the effective lower bound? Let’s look at some data on policy rates shown in Figure 3. I have split the countries around the euro area into three groups and taken simple averages of their policy rates. The green line is the ECB MRO rate and the brown line is the EONIA rate. From the EONIA rate, you get a sense of what is going on with the effective euro area overnight rate. The blue line at the top is Southeastern Europe, the red line is Central Europe, the pink line is average of Switzerland, Sweden, and the United Kingdom. If you look at the average interest rate spread between the euro area and these different groups, in Southeastern Europe the spread was approximately 3 percentage points in 2007–2008 and that’s pretty much what it is today. If you look at the average of the Central European countries, it was about 1 percent in 2007, and is approximately 1 percent today. And if you look at the other countries in Europe, 0 percent in 2007 is pretty much an average 0 percent today.
This then raises the question: has anything really changed with negative euro area rates given that the spread to the euro area has not changed very much? It certainly creates problems for reserves management in central banks, so it is appropriate that the conference covered this topic on the first day. But, in terms of monetary policy, it seems that for most economies there really hasn’t been anything systematically new about the spillovers from negative euro area rates in terms of the policy response of other monetary authorities.
However, moving beyond these averages, there are two countries that appear to have been affected by the lower bound constraint, both of which have historically had, or, at least at the beginning of the crisis, tended to have, rates that were below those of the euro area: the first is Switzerland. In 2007, the spread between the euro area rate and the Swiss Libor rate was −1.5 percentage points, today the spread is approximately −0.4. They have been unable to maintain this historical wedge with the euro area rate. Another country that had policy rates below the euro area before the crisis was the Czech Republic. Their spread was about −1 percent before the crisis and now they have a positive spread of 0.4 percent. This is evidenced by Figure 4. Perhaps it is no surprise that these two countries are the ones that have used exchange rate policies to try to maintain an effective monetary policy spread between the euro area rates and their own rates in order to offset exchange rate appreciation pressures.
But, when you look at Southern Europe – here I am talking about countries where they have floating exchange rates – there really hasn’t been much of a change. Even when you look at the minimum of that group in the region, they still seem to have this headroom above the zero lower bound.
Does the heterogeneity in the monetary policy space caused by negative euro area rates affect any outcomes?
I ran the very simple regression shown in Figure 5; it is not in any way causal, but I wanted to use it to consider the costs of deflation, given that some countries may be facing more deflationary pressures than others due to exchange rate appreciation. In the regression, next year’s inflation expectations from Consensus Economics are regressed on current inflation, a dummy variable indicating whether the country is in deflation or not, a dummy variable indicating when the ECB has negative interest rates, and the interaction between this deflation dummy and a variable indicating when the ECB has negative interest rates.
The coefficient on the ECBnegt term shows that the ECB cut rates into negative territory, was associated with an increase in inflation expectations unless the country was in deflation. However, when an economy was already in deflation, negative ECB rates were associated with more deflationary pressures. Thus, for some countries, negative euro area rates may have potentially exacerbated deflationary pressures. What sources of costs may come from deflation if you can’t cut rates enough to offset exchange rate appreciation pressures at the effective lower bound? So far, they have not been associated with very bad consequences such as deflationary spirals.
But there might still be other costs. Figure 6 above plots the current inflation rate of a country against the dispersion in next year’s inflation forecasts from the consensus panel of economists. What you see is a U-shape relationship. The right tail is a well-known result: when inflation is higher, there is greater dispersion in views on inflation and potentially a greater dispersion in prices. In New Keynesian models, the dispersion of prices is the source of costs from higher inflation as they cause an inefficiently allocation of resources. What this graph shows – which is new – is that there is a similar rise in dispersion as you get into deflation territory; it is almost symmetric. One of the costs of deflation could well be this increased dispersion in inflation expectations.
But an economy can go into deflation for different reasons. For example, it can go into deflation because the exchange rate appreciated, or because of deleveraging pressures from high debt, which reduces demand. What we do here in Figure 7 is to look at whether the shock, causing deflation, and the monetary policy framework affect the dispersion of views on inflation during deflations. From the top line, it says that if you are in an inflation targeting regime, then there is a significant, positive relationship between deflations associated with high debt and the dispersion in inflation expectations.
However, if you are an inflation targeter, movements in the exchange rate do not have much of an effect on dispersion. But, the story is a little bit different if you are not an inflation targeter. Non-inflation targeters experience a significant increase in the dispersion of inflation expectations when exchange rates appreciate. This suggests that as some countries are potentially moving or changing their frameworks away from trying to strongly meet inflation targets over a short horizon, this may increase the dispersion in inflation expectations if the economy is experiencing deflation.
Now let me move on to spillovers through the banking sector and how they may change with negative euro area rates. The last two panels of Figure 8 show the effect of monetary spillovers on cross-border flows, which we explained by financial frictions. In the paper, we have two financial frictions; one is the double banking friction I explained earlier, and the other is that some liabilities in emerging economies may be in foreign currency.
When interest rates are cut in the center country, it boosts the net worth of banks in the center country. By boosting the center country banks’ net worth, this boosts capital inflows into emerging markets. If you have a lot of FX debt, then the currency appreciation also boosts the net worth of emerging market banks, as the value of their liability falls relative to the value of their assets. This then causes a compression in emerging market spreads, which boosts borrowing and growth.
The effect of foreign currency liabilities on growth has been recently discussed in the BIS annual report. You can construct effective exchange rates series in many ways, as shown in Figure 9. It can be based on trade flows using imports and exports. This is the traditional way of measuring an effective exchange rate. Some of my colleagues at the BIS have instead recently constructed effective exchange rates based on the currency composition of an economy’s debts, a liability-weighted exchange rate. For both advanced economies and emerging economies when there is appreciation of the trade weighted exchange rate, this is negative for growth. This is the standard competitiveness channel. But there is a special channel that only seems to operate in emerging economies, which works in the opposite direction. When the debt-weighted exchange rate appreciates because the value of liabilities falls relative to assets – it actually boosts growth in emerging economies. This can explain why the spread is somewhat more volatile in emerging economies than in advanced economies. So the movement into negative euro area rates, while it may be negative for the growth of countries like Switzerland, Sweden, and the United Kingdom due to currency appreciation, it may be more positive for emerging economies that have foreign exchange liabilities.
Negative rates squeeze euro area bank profitability
Negative rates squeeze euro area bank profitability
Negative rates squeeze euro area bank profitability
We have seen some of these exchange rate appreciation pressures, so that would result in positive spillovers from the euro area to the region. However, if negative rates impair bank profitability, our model suggests these spillovers might be weaker. Here, as shown in figure 10, I have plotted the equity price index for euro area banks, that’s the blue line, and the equity price index for the entire index in the euro area, which is the red line. The two black vertical lines are the points when the ECB cut interest rates into negative territory and when they reduced them further. What you see is that around these announcements, bank equity prices fell relative to the overall market. So, perhaps different from past spillovers, where cuts to policy rates have boosted the net worth of banks, negative interest rates seem to have been associated with a decline in the net worth of euro area banks. In our model, this would imply that negative rates would have a weaker effect on cross-border flows than normal interest rate cuts. As banks typically intermediate credit to households and to firms, spillovers to these sectors are potentially weaker with negative euro area interest rates. But, a lower capacity of euro area banks to intermediate capital flows may mean greater cross-border intermediation through capital markets. So, with the movement into negative rates, there could be changes in the way spillovers work, depending on the depth of a country’s capital markets.
Lower risk premia following negative euro area rates
Lower risk premia following negative euro area rates
Lower risk premia following negative euro area rates
For many countries in the region, the government is the only major issuer of bonds, so capital flows caused by negative rates will tend to go into the government bond market rather than the nongovernment sector. Indeed, there is some evidence of such spillovers to government bond markets in the region. Following ECB policies, which pushed interest rates into negative territory, CDS premia on sovereign debt have declined in Central Europe and Southeastern Europe. This suggests that negative rates reduced the risk premia on sovereigns in both Central Europe and Southeastern Europe, but had little effect on the other European economies.
To conclude, does anything change with negative euro area rates? I highlighted potentially two channels. The first channel is that some countries may hit the effective lower bound. This is more likely in countries with historical interest rates close to the euro area rates. But, if a country had high rates relative to the euro area, it is likely that there would be little difference in spillovers from interest rate differentials. In terms of the effect of negative rates on bank profitability, if there were an adverse effect of negative rates on euro area bank profitability, it would weaken the spillovers through cross-border flows that are intermediated through banks, weakening spillovers to the non-government sector, but potentially increasing spillovers that come through capital markets.
