Will Commodities Deflation Save the Brazilian Central Bank?

The most recent inflation data points for Brazil are a bit encouraging in the sense that some moderation might be working through the system, mainly due to better behaved increases in food prices. More interesting, the first preview of the FGV’s IGP-M index brought a negative reading in August pushed by a deflation in the WPI (or IPA-M) a direct consequence of the downward correction in commodity prices seen since mid-July. We wonder how causality works on a simplified but well specified system containing key inflation measures as well as changes in the foreign exchange rate, monetary policy rate (Selic rate) and the CRB commodities index. Our question is whether there will be a pass-through to the official IPCA index and if this will ultimately make the BCB’s life easier.

1) Causality

We model two VARs (Vector Autoregressive) using the following variables: IPCA inflation, IPCA food inflation, IPA-M (wholesale prices index by FGV) inflation, the percentage change in the foreign exchange rate, the Selic rate and the percentage change in the CRB commodity index. VAR1 was run using the m/m percentage changes of the above variables while VAR2 uses the y/y percentage changes. After finding the optimum lag structure both models seem to be well specified and no autocorrelation in the residuals were found. We estimate two versions in order to see if seasonal effects and short-term noise, commonly present on the m/m data could have any impacts on our conclusions.

We find the following chain of causality. Results were the same for both VAR1 and VAR2:


As we can see from the simplified diagram above, IPA-M is affected by both past changes in the currency (BRL) and the CRB. Going forward, IPA-M changes will impact both the IPCA index directly and the food and beverage sub-category of the IPCA index (IPCA Food). We can also see a statistically significant causality from changes in the exchange rate into the IPCA food as well as into the IPCA index. We do not present the Selic rate causality arrows in the diagram above to make it less complicated. Although, the results are sound in the sense that IPA-M, IPCA Food, IPCA and BRL all granger cause the Selic rate, meaning the monetary authority do respond to movements on those variables. The causality between the Selic rate and the IPCA is actually bi-directional while the CRB does not granger cause the Selic directly. Quite reasonable since the CRB is supposed to represent the exogenous variable in our system.

In other words, the recent downward correction in commodity prices will likely push the IPA-M inflation to the downside, which in turn will put disinflation pressures over IPCA food as well as the IPCA index.

2) Impulse Shock Analysis

A quite interesting exercise is to take a look at the impulse responses to a shock in both CRB and IPA-M. Even though the bands nearly contain the zero line in every response to a CRB innovation (first line of charts), we should say our bands here are not quite free of heteroskedasticity anomalies. In this sense, the pattern of the solid line could be a better indication of the responses. Therefore, a shock on the CRB does produce a quite concentrated response by IPA-M for at least the first twelve months. The responses of IPCA food and the IPCA index are 13 and 15 months long when the response becomes inversely proportional to the shock, intuitively explained by the fact that the BCB might act against such a long-lived shock response. image004_512_01.png

On the meantime, the responses of an innovation on the IPA-M index by both IPCA food and IPCA index are more well defined as the lower bands actually remain above the zero line (significance level) for five and eight months respectively.

Summing up, a shock on the CRB index indicates that there will be a direct response by all the inflation indexes we analyze here. While it seems more pronounced for the IPA-M, it is also positive but not clearly significant for the IPCA food and IPCA index. On the other hand, the direct response of a shock on the IPA-M (which could arise from other sources than commodities – semi-industrialized inputs for example) is clearly significant and positive for five and eight months in terms of IPCA food and IPCA index responses respectively.

3) Exchange Rate Remarks

An interesting point is whether the collapse in commodities will mean BRL depreciation and if this effect will actually compensate the movement seen in commodities. Well, the real has lost some ground recently but on a much smaller magnitude. So far in August it has lost 3.5% against the dollar vis-à-vis a 7.0% drop on the CRB. Even assuming a more dovish BCB policy over the next months, which is unlikely, the interest differential remain quite high and the country should continue to receive important USD flows. In this sense, we see little scope for sharp depreciations, while a stable or appreciating real would not surprise us.

4) Final Conclusions:

The main conclusion from this study is that the recent fall in commodity prices, measured here by the CRB is likely to put downward pressures on the targeted inflation index IPCA. The CRB index fell 10% in July and so far this month is down by another 6.8%. Therefore, we might see the IPCA inflation benefitting from this recent drop in commodity prices, although the effects might take a couple of months to work through the system. But, is the BCB’s life easier after the fall in commodities? It really depends on how sustained is the fall and also on the fact that no rebound is seen on the near future. If we assume the recent fall in commodity prices is based on weaker fundamentals of the global economy, it might well be the case that prices will remain on the downside and also that overall semi and manufactured inputs will also see downward pressures, which will certainly push inflation to the downside in Brazil. However, the BCB is unlikely to change the pace of rate hikes in the near future. The high levels of capacity utilization still threat the overall balance between aggregate demand and aggregate supply. We think the commodities effect is positive but it’s too early to change our COPOM call, which is set for another 75 bps hike on the September meeting.

3 Responses to "Will Commodities Deflation Save the Brazilian Central Bank?"

  1. Monica B. de Bolle   August 13, 2008 at 12:31 pm

    There´s strong evidence that more recently the BRL has responded not only to widening interest rate differentials, but also to robust growth in export prices. At the same time, ever since Meese and Rogoff (1983, we know that forecasting exchange rate movements in the short/medium term is a notoriously difficult exercise, not least because relationships between the e-rate and its macro fundamentals shift over time. This said, movements in export prices have recently performed quite well in anticipating shifts in the nominal e-rate. If that holds true, then the expected fall in export prices as a result of the fall in commodities will certainly exert some depreciating pressure on the exchange rate. And that is bad news for inflation in the near term. No wonder then that the yield curves in Brazil have not budged, despite the “improved” inflation numbers.

  2. Anonymous   August 14, 2008 at 4:58 am

    Hi, thanks for the analysis.The CRB is rightly assumed exogenous, so should it not be that the SELIC (and all the other variables in the system) does not Granger cause the CRB and not “the CRB does not granger cause the Selic directly” as mentioned above? Perhaps my confusion.

  3. Italo Lombardi   August 14, 2008 at 8:31 am

    Thanks a lot for all the comments.@Monica Bolle:Indeed, forecasting FX movements in the near term is quite a challenge. We may have some inflationary pass-through in case the currency depreciates but it looks like the lag between this depreciation and impacts over the inflation is bigger than the lag between commodities and inflation figures. Although, the FX do impact inflation expectations quite heavily – very interesting balance of forces working here.@Anonymous:Well, if central banks were really to react only to inflationary pressures caused by an excessive real money supply (vis-a-vis the demand for money), than an exogenous influence from a commodity shock should not Granger cause the Selic too. More or less the rationale why CB should all focus on core measures of inflationAlthough the opposite (as mentioned by you) is of course true – “SELIC (and all the other variables in the system) does not Granger cause the CRB” and we are glad that they don’t!Italo Lombardi