Tuesday, July 15, 2014

Monetary Policy: Canada and the United States

Since the financial crisis, monetary policy has been strikingly different in Canada and the U.S. The first chart shows overnight nominal interest rates - the overnight money market rate in Canada and fed funds rate in the U.S.
The fed funds rate has been below 0.25% since late 2008, but the Bank of Canada achieved liftoff in mid-2010, and the overnight rate in Canada has been at 1% since then. What about central bank balance sheets?
In Canada the size of the balance sheet increased modestly during the last recession, then declined somewhat. Currently total assets held by the Bank of Canada are slightly less than 5% of GDP. But, as is well-known, successive rounds of quantitative easing (QE) have increased the size of the Fed's balance sheet substantially, to close to 25% of GDP. As is also well-known, QE has lengthened the average maturity of the assets on the Fed's balance sheet. Currently, about 2/3 of the Fed's Treasury security holdings exceed 5 years to maturity, and the Fed holds a large quantity of long-maturity mortgage-backed securities. The next chart shows the composition of the Bank of Canada's portfolio of Canadian government debt.
In contrast to the Fed, which has reduced its holdings of Treasury bills to zero, the Bank of Canada still holds a large fraction of its assets as T-bills, though this fraction has come down since before the financial crisis. However, most of the shift in the Bank of Canada's portfolio was out of T-bills and into government bonds with maturity less than three years. Before the financial crisis, the Bank held about 2/3 of its portfolio of government debt in maturities less than three years, and that is roughly the case currently.

So, since the end of 2007, the Bank of Canada has increased the size of its balance sheet by a small amount, and lengthened the maturity of its government securities, also by a very small amount. There's really not much going on relative to the large QE intervention that occurred in the U.S., which included the purchase of a large quantity of asset-backed securities. Thus, the conventional view, given the currently higher overnight nominal interest rate in Canada, would be that Canadian monetary policy is substantially tighter than in the U.S. So, if I were a strong believer in the persistent real effects of monetary policy, if I thought that lower nominal interest rates meant higher inflation, and if I think that QE works as advertised, I might think that: (i) real economic activity should be depressed in Canada relative to the U.S., and (ii) inflation should be lower in Canada than in the the U.S. Is that what happened? First, real GDP:
The U.S. lost some ground to the U.S. during the recession, but growth in the recovery phase has been roughly similar. Next, the PCE deflator:
Again, there is a level effect in Canada vs. the U.S., relative to 2007Q4, but in the recovery phase the PCE inflation rate is roughly the same in the two countries, though somewhat higher in Canada in the last couple of years.

So, if you were to ask your average macroeconomist to back out monetary policy in Canada and the U.S. by looking at the last two charts, that person might tell you that it was about the same. But we know it wasn't.

I have seen a lot of stories recently about the effects of monetary policy on asset prices - bubbly talk, basically. If we take those stories seriously, we might expect to see more asset value appreciation in the U.S. than in Canada. In the stock market, that is certainly the case. The next chart shows the S&P 500 index, and a comparable measure for Canada.
But, not so for the housing market (these house price measures are not really comparable - Case/Shiller compared to a new house price index for Canada; but that's the best I could do):

So, maybe monetary policy - conventional or unconventional - isn't as big a deal as some people think it is.

22 comments:

  1. "So, if you were to ask your average macroeconomist to back out monetary policy in Canada and the U.S. by looking at the last two charts, that person might tell you that it was about the same. But we know it wasn't."

    The average macroeconomist who thins he can back out the stance of the monetary policy from those two charts should really reconsider his training in macro. In fact if that macroeconomist had taken a look at the last two charts and observe that while inflation rate and gdp growth rate have been very similar, asset prices have behaved very differently and hence maybe the monetary policy in the two countries have been very different! And maybe this difference in the behavior of asset prices (which could potentially be due to the differences in monetary policy) has led to a similar behavior in real activity, without it the gpp and inflation taking completely different paths.

    ReplyDelete
    Replies
    1. Yes, of course the average macroeconomist could be confused.

      Delete
  2. Steve: Here's the Canadian version of Case/Shiller: http://www.housepriceindex.ca/

    It doesn't change your story much.

    ReplyDelete
  3. Steve: Here's the Canadian version of Case/Shiller (Teranet/National Bank): http://www.housepriceindex.ca/

    It doesn't change your results.

    ReplyDelete
  4. Sorry for the double-post above. Computer glitch.

    A Market Monetarist would look at your graphs for real GDP and the price level, and say: "Look! Nominal GDP has been rising faster in Canada than the US, which means that Canadian monetary policy has been looser than US monetary policy".

    And he would then look at the other graphs and say: "And this has caused nominal interest rates to be higher in Canada, asset prices to be higher, and the central bank's balance sheet relative to the size of the economy to be smaller. Just as we predicted."

    But if you think of monetary policy as setting interest rates, as many do, then you have to think about a system like an inverted pendulum, where if you want to move the pendulum to the North, you must first move it South, then move North as it leans North, to stop it falling over.

    Except that economies, unlike inverted pendulums, have expectations. It's the *threat* to move interest rates South that causes them to move North. In equilibrium, if the threat is credible, and you announce your target clearly, you don't need to carry out your threat.

    ReplyDelete
    Replies
    1. I read this post thinking I had to tell an MMist, so they could make the very comment Nick makes here. :D
      Scott Sumner too makes a good point in a recent post by going to the extreme case of hyperinflation:

      "just consider a monetary policy shock that no one could question—hyperinflation. Wheelbarrows full of billion mark currency notes. Can we all agree that that would be “easy money?” Good. We also know that hyperinflation leads to extremely high interest rates and extremely low real cash balances .... In contrast, Milton Friedman would tell you that really tight money leads to low interest rates and large real cash balances, exactly what we do see."

      But at the same time, exactly how tight or loose is monetary policy being made by expectations and CB communication effectiveness? Is it 7? Or 34? What should the measure of tightness be in order for monetary policy to be able to do it's job? 100? 10? Can we predict what it will be in five years? Isn't this all just too conveniently unmeasurable and unknowable? It seems there's no way to independently measure tightness or expectations or CB credibility (outside of the above circle of logic) so that we can test this hypothesis. We could just as well substitute a phlogiston hypothesis: OMOs and QE are not sufficient on their own: they must be accompanied by sufficient phlogiston to work properly. How do we know if we have enough phlogiston? Only one way to tell: OMOs and QE will work as expected with sufficient phlogiston, and they won't when phlogiston levels are too low. Show me a CB struggling to hit its inflation target and I'll show you an economy lacking phlogiston. Canada clearly has sufficient phlogiston levels while the US does not. Thus the phlogiston hypothesis is 100% in agreement with all empirical data (and 100% useless too).

      Wouldn't it be nice to instead have a means of independently calculating a quantitative measure that explained why some economies react one way to concrete CB actions, and others react another way? That way the hypothesis could be tested by calculating this measure independent of checking whether or not monetary policy is working. Also predictions could be made regarding economies not yet in need of unconventional monetary policy. I know of at least one existing example of such a hypothesis. Is it the best example? I don't know, but (unlike phlogiston & related ideas) at least we can put it to the test!

      Delete
  5. Nick Rowe's comment, above, is tautological. How do we know monetary policy has been looser in Canada? Because NGDP rose faster. Why did NGDP rise faster? Because monetary policy was looser. Like any tautology, it may be true, but it is also unhelpful.

    ReplyDelete
    Replies
    1. Market monetarists tend to point to how markets react to information about the stance/actions of the central bank as evidence. So when the Bank of Japan recently embarked on Abenomics, or Switzerland switched to targetting the exchange rate, markets reacted as if monetary policy mattered.

      Delete
    2. I don't think that argument is particular to market monetarists. People use event studies to argue that QE mattered.

      Delete
    3. Nick Rowe's comments are almost always unhelpful.

      Delete
  6. canada: smaller shock, smaller monetary policy response. US: bigger shock, bigger monetary policy response. We dont have counter factuals, so how are these pictures in any way challenging the efficacy of monetary policy?

    ReplyDelete
    Replies
    1. In this business, we never have counterfactuals. How do you know it was a bigger shock? What was the shock? Isn't North America tied together closely, in terms of both real and financial interaction? Maybe Washington state and British Columbia have more in common than Washington state and Florida? Except Washington state and British Columbia are subject to different monetary policy.

      Delete
    2. Housing prices didn't crash in Canada. I believe bank regulations were different, and Canadian banks were better capitalized and had less exposure to mortgage debt. I don't think Canada experienced a financial crisis and credit crunch at all like the US. Consumer debt overhang and similar issues were less in Canada. This all points to a smaller shock.

      Delete
    3. Exactly. Some of the comments above baffle me. The financial network system of the US and Canada couldn't be more different. Not to mention the whole corridor versus overnight targeting thingy.

      Delete
  7. we know from the extensive literature on home bias (see for instance McCallum 1995 for the specific case of US vs Canada) that sovereign borders matter for trade and financial linkages. So although washington is closer to BC, it probably comoves more with florida. I agree with you that since shocks are unobserved, its hard to determine whether the “shock” that hit north america was more severe on the US or Canada. All i am saying is that given that it could be the case, these pictures don't tell much about the efficacy of monetary policy ( If i were to argue against the efficacy of monetary policy, i would rather point to VAR evidence from the last 2 decades which seems to suggest that monetary policy shocks have had a diminished impact on macroeconomic variables compared to the past)

    ReplyDelete
  8. I guess it might be good to cite the data sources to pay credit to them...

    ReplyDelete
  9. More Japan data is in. It seems to challenge your previous post. I updated your chart here:

    http://badoutcomes.blogspot.com/2014/07/japan-update-weve-waited-lets-see.html

    ReplyDelete
  10. re: US & Canada: can someone here produce a three-parameter or fewer theory based model of the general price level and/or inflation that performs at least this well as this model does against these two Fed models and against TIPS spreads? [1], [2], [3], [4], and which you're willing to stack up against its predictions for Canada for late 2015 and for the US over the next six years (out to 2020)? I'm not the author of the model, but I'd love to see what other people have to compare with it. Any takers?

    ReplyDelete
    Replies
    1. Go away, and take that crackpot with you.

      Delete
    2. What's crackpot about it? I'm actually curious ... the results essentially are identical to Paul Krugman's view. Maybe you think Krugman is a crackpot; that would help me gauge my ... um, crackpottiness.

      Delete
  11. This post neglects the elephant in the room, the housing bubble in the US. Taking this shock into account the last graph clearly shows that after the housing bubble burst QE lead to a sharper increase in housing prices in the US than in Canada. Of course this was only a side-effect of QE, the main purpose of purchasing private assets (CE buying government debt aims at something else) is to help private players in deleveraging. I know that some economists think that the structure of debt is irrelevant and in general I would agree but in the case of balance sheet recessions like the Great and the Little Depression we have to take Fisher, Minsky and Koo seriously.

    ReplyDelete
  12. Steve, on a different note, Cochrane has a post on the Overnight Reserve Repurchase Facility. Given our discussions and your recent published work, I assume you also favor the initiative. Do you think it can ameliorate the scarcity in short-term bonds and help facilitate transactions?

    ReplyDelete