10 September 2007

Growth, Inflation and Globalization - Reviving the debate on causes of inflation

In June, Brazil’s Central Bank changed its forecasts for inflation and GDP growth. The latter was increased from 4.1% to 4.7% for 2007, which ceteris paribus should imply lower unemployment and thus higher inflation when compared to the previous forecasts, but interestingly enough, expected CPI inflation was reduced from 3.8% to 3.5% for the same year! In Bloomberg’s piece of news reporting this contradiction, it’s argued that the “accelerating economic growth in Latin America's biggest economy isn't sparking inflation as a currency rally slashes the cost of importing goods”, but can this be true?

Estimates of exchange rate pass-through on inflation show a modest effect when the home currency is depreciating. When the currency appreciates, as in this case, downward price inflexibility further reduces the effect import prices have on the aggregate price level, making it implausible the argument that the Real’s higher value has a stronger effect over inflation than the business cycle dynamics.

A solid explanation must show how the determinants of inflation may shift and lead to this unlikely combination, in other words, must show why the ceteris paribus assumption doesn’t hold. Because the goal here isn’t to discuss the Brazilian economy, we shall leave aside other considerations and show that the main and ultimate reason for the inflation to fall as the GDP rises is in this case a positive supply shock brought about by globalization.

For the higher GDP to be consistent with lower inflation, one of variables usually taken as constant must be changing, for example, if the potential growth were to increase by more than the expected actual growth, then inflation should fall. The only problem is that potential output depends on the structure of the economy and it makes no sense to think a priori it increased.

Alternatively, if we consider an augmented Phillips Curve, one of the other determinants of inflation may be pushing it down by more than the output gap’s upward pressure. This could be the case of a reverse oil shock – but the oil prices are on the rise –, increased central bank credibility – but that still doesn’t account for the higher GDP growth –, or even currency appreciation – but this isn’t enough to outweigh the business cycle pressure, as said above. It could also a positive productivity shock, however the last we saw was the personal computers revolution in the 80s. Or was it?

As presented in the World Economic Outlook (2006), globalization increases “pressures to innovate and other forms of nonprice competition”, raising productivity growth. This allows higher output without accelerating inflation. If the Central Bank recognizes the positive supply shock ahead of the private sector, “it can take advantage of its better forecasting to opportunistically lower inflation while delivering output growth rates that pleasantly surprise the private sector”. This is what Rogoff (2006) calls “opportunistic disinflation” and perfectly fits the Brazilian case.

Such explanation implies a permanent effect from globalization on inflation, for the Central Bank now targets a lower inflation rate – despite the official government target of 4.5% for 2008, the monetary authority already announced it will actually aim at 4%. Of course there’s no free lunch and the country gave up the opportunity to have an even larger GDP growth without accelerating inflation.

Because there’s general consensus that the long-run price level is determined by the money supply, economists agree that monetary policy would have to be altered for globalization to impact the inflation trend. And we have just shown how it can do so by creating policy incentives. Another way globalization can affect inflation is temporary shocks that change its short-run behavior, but theory for explaining it is not as widely accepted. For instance, Ball (2006) says “applied economist typically analyze short-run inflation behavior with a Phillips Curve”, but what to include varies a lot. In his paper he estimates 2 specifications and mentions a few other possible modifications.

In both cases he finds little to no effect: when testing the relevance of foreign output gap, he “pooled annual data” for a number of countries and found it “barely significant” and “at most a secondary influence on inflation”; when testing the role of trade, it had “at most a small effect”.

The problem is such results are contrary to the most prominent literature on the theme and his methodology way too simplistic to deal with the econometric difficulties. When assessing the role of trade, our pooled OLS estimates also resulted in small and statistically insignificant coefficients for openness, but tests on fixed and random effects strongly indicated that simultaneity was an issue that had to be addressed. Using instrumental variables to control for it, openness came with the expected sign and statistically significant.

As for the theoretical arguments, it may be useful to analyze Mankiw’s comment on Ball’s article for a Fed meeting. They both agree it’s possible competition may have lowered the “typical markup of price over marginal cost” and provided a “one-off beneficial shock to the inflation process”, but what really matters for the cyclical behavior of inflation is the sensitivity of mark-up to the business cycle.

Mankiw explains that very well, however both of them take for granted that “firms face diminishing returns as their output expands”. This neglects that the best reason for firms resort to outsourcing is to exploit economies of scale, and ignoring that is ignoring an essential feature of today’s globalization. In such a world, markups may still not be countercyclical, but part of firms’ desired prices is. Firms’ average costs go down as their production increases, for they benefit from EOS, and, ceteris paribus, prices fall. Of course prices don’t actually have to fall, firstly because markups may be acting on the opposite direction and secondly because actual prices also depend on the rate of adjustment. The important thing is globalization can put pressures on prices, dampening or compounding with other determinants. The more firms outsource, for example, the greater benefits they reap from EOS and the larger the impact on price dynamics.

We have thus a way for globalization to change the inflation process by affecting desired prices, in line with models of trade with imperfect competition and scale economies, and there is evidence from sound econometrics corroborating it [See WEO (2006)]. Additionally, one can consider the direct role of import prices and the exchange rate, which has been shown to have a marginal yet significant effect. Finally, as argued in the literature, globalization provides policy incentives that can alter long-term inflation trends, which fits the Brazilian recent development.

The extent to which these channels can alter the domestic price level is yet to be measured in consensus, but it is harder and harder to claim globalization is not crucial for the analysis of inflation.

-- André Luis Pulcherio



Anonymous said...

Excellent article... Is it a part of a bigger paper or sth?

ALPulcherio said...

Thanks, anonymous. As a matter of fact, it is largely inspired in the paper where I made the analyses. I took the main points and avoided the econometrics here.