We have been very active in our discussions of the impact of the pending rise in food prices around the world (from central bank largesse to weather-related chaos). As Goldman notes, food inflation has been one of the most significant sources of headline inflation variation in emerging markets (EM) over the past few years. Since June, international prices for agricultural commodities have risen almost 30%, increasing the risk of fresh, food-related increases to EM headline inflation. We, like Goldman, expect EM headline inflation to start to reflect the relevant pressures more broadly in the October prints at the latest. While the effects, for now, are expected to be less extreme than the 2010-2011 episode, the timing as the US enters its fiscal-cliff-prone malaise, could mean a further round of easing will reignite this critical inflationary concern.
Via Goldman Sachs, Food prices: A key driver of EM inflation
Swings in food prices have important implications for overall inflation in emerging markets. Since 2007, we have observed substantial shifts in food inflation, which in turn have triggered significant contemporaneous volatility in EM headline inflation (see Exhibit 1).
Food inflation has a strong impact on overall EM inflation for two reasons:
- In lower per-capita GDP economies, households necessarily dedicate a larger portion of their disposable income to inelastic goods such as food. As such, food makes up a larger share of the consumer basket. The average inflation share for food items in EMs is generally larger than that for the G10 countries (25% vs 15% respectively, on average). In order to capture the joint effect of the weight, the relative variation of food vs non-food inflation and the potential correlation between food and non-food items, we run univariate regressions of food on headline inflation. The R-squareds are typically higher on average for EMs (42%) than for G10 economies (33% respectively, Exhibit 2).
- Food prices have been highly volatile since 2007 globally. We have observed very large spikes in international prices for agricultural commodities (proxied by the S&P GSCI® Agricultural Index) in 2008, 2011 and more recently in June 2012. Such global price shifts typically also tend to be reflected in local food inflation. Exhibit 3 shows the co-movement between international food prices and an equally weighted average of food inflation rates across emerging markets. International food prices have tended to lead local food inflation by a few months (approximately four months on average).
Following a significant increase in 2010, aggregate EM food inflation peaked in 2011 and has contributed to an overall moderation in EM headline inflation since. But EM food inflation has recently shown tentative signs of a trough and, at the country level, there is variation in the recent path of food inflation. China, Korea and Indonesia have seen the largest falls in food inflation from their 2011 peak. However, in countries such as Taiwan, Mexico and the Czech Republic, yoy food inflation has picked up and is currently hovering at higher levels than in 2011.
This bottoming-out of EM food inflation has coincided with a significant spike in international agricultural commodity prices. In June and July this year, the S&P GSCI® Agricultural Index rose almost 40%, to levels last seen in August 2011, and roughly speaking has remained there since. Should this spike persist, we would expect to see food inflation pick up across EM once again.
Here we argue that food price pressures will boost EM headline inflation by October at the latest. However, we do not expect EM CPI to exceed 2011 levels (in yoy terms). This is because we expect the increase in food prices to be smaller and less broad-based, and because non-food inflation is running at a slower pace currently. Moreover, we find evidence that the pass-through from international to local food prices has declined, something that first became visible in 2010.
Food price outlook – new highs expected
Agricultural commodity prices have exhibited substantial swings in the past few years. On the demand side, rapid income growth in EM economies has supported overall demand for agricultural products. Along with the broader increase in agricultural commodity demand, increased consumption of meat products has led to higher meat production and, in turn, higher demand for livestock feed. Lastly, high energy prices also boost food demand via the substitution process between conventional fuel and biofuel.
Given this backdrop of elevated demand for agricultural commodities, the response in food supply conditions becomes the key to analysing price movements. Volatility in weather patterns and crops has helped trigger substantial inventory shortages and price spikes such as those experienced in 2008, 2011 and more recently in June 2012.
The current spike has come in response to the summer drought in the US Midwest, which was one of the worst in the past century. In addition, a wide set of agricultural commodity producing countries have experienced adverse weather conditions (such as Brazil and Argentina in the past winter, and Russia, Ukraine, Kazakhstan and India). Damien Courvalin from our Commodities Strategy Team points out that these disruptions have caused substantial losses in global food supply (see Agriculture Update: ‘Severe US Drought to Push Corn and Soybean Prices to New Highs’, July 23, 2012).
The supply loss is concentrated in wheat, corn and soybeans, which jointly account for 70% of world agricultural production. In contrast, rice remains largely unaffected.
Despite the resulting 40% spike in the S&P GSCI® Agricultural Index between mid-June and mid-July, demand for agricultural commodities has remained robust. The net result has been a decline in inventories, with the USDA’s September 1 stocks of corn and wheat well below expectations, as Damien highlights in Agriculture Update: ‘Crop prices to recover on tight supplies with corn outperforming’, September 30, 2012.
Our Commodities Strategy team expect demand to remain resilient and supply to remain binding, leading soybean and corn prices to new highs in the coming months. Higher prices will eventually be followed by a supply response, and if weather returns to normal, we should expect a large crop in South America (harvested next spring) and in the US (harvested next autumn). In the interim, prices are likely to remain high.
However, there is a clear weather dependency to this assessment; further weather adversity is likely to pose further upside risks to food prices. To address the binary nature of the food price outlook, our Commodities Strategy team provided us with two scenarios:
- The ‘favourable’ weather scenario, in which larger harvests in South America and the US serve to moderate agricultural prices following the initial increase. In this scenario, a basket of corn, wheat and soybeans sees year-on-year price changes of 46%, 16% and -21% in 3, 6 and 12 months respectively.
- The ‘moderately adverse’ weather scenario, in which supply tightness intensifies due to less favourable weather in South America, pushing prices to a higher peak over the coming months. In this scenario, the basket of corn, wheat and soybeans increases 65%, 41% and 1% in 3, 6 and 12 months respectively.
Exhibit 4 shows the equivalent paths corresponding to each of the two scenarios of price developments in the corn, wheat and soy basket. In both scenarios, the S&P GSCI® Agricultural Index reaches new highs in the months ahead and declines one year out. The peak is, of course, higher in the adverse scenario, as is the trough 12 months out. The decline following the initial spike is also more gradual in the adverse scenario, while the final levels remain very close to the previous (2011) highs. It is worth pointing out that this scenario analysis is only meant as an illustration of the broader argument, rather than a precise forecasting exercise.
Evidence of a moderation in the pass-through to EM inflation
To translate our scenarios for international food prices into local food price trends for emerging markets, we need an estimate of the relationship between the two variables. As mentioned earlier, large shifts in global food prices have tended to show up systematically in local food inflation. Moreover, local food prices are typically stickier and slower to respond to shocks in global agricultural prices, which creates a lag between the two.
To map international food prices onto local food prices, we follow the framework we introduced in Global Economics Weekly 11/13, June 6, 2011. We regress changes in the S&P GSCI® Agricultural Index on changes in an equally weighted average of food CPI components from key EMs. To avoid issues of seasonality and excessive near-term volatility, we look at year-over-year percentage changes in the two variables. Lastly, we examine different lags in international food prices to find the type of structure that offers the highest explanatory power. As in our previous analysis, we find a strong correlation between international and local food prices (an R-squared of 40%), with international food prices feeding through to local food prices with the highest explanatory power at a four-month lag (with a five-month lag a very close second).
We estimate the historical sensitivity of local to international food prices at around 0.058, which implies that a 10ppt increase in international food prices would tend to raise our proxy of EM local food inflation by 58bp. Interestingly, this is 20% lower than our estimate from one year ago, of 0.073. This is further evidence for our suggestion from last year that EM CPIs appear to be displaying a lower sensitivity to global food price shocks. This could be due to a number of reasons, such as the temporary nature of the shocks, the softening in global demand dynamics leading to less broad-based price pressures, or the larger capacity of EM authorities to respond to food price volatility and smooth such shocks. It will be interesting to observe whether the pass-through declines further this time too.
In our previous analysis, we also examined two alternative scenarios for food prices: one that assumed that normal weather conditions persist and one that assumed that adverse weather conditions push food items significantly higher. Based on those scenarios (combined with our pass-through estimates), we projected ranges of outcomes for the forward path of our EM food inflation aggregate. Finally, we translated those paths into EM headline inflation projections by keeping the rate of inflation for non-food CPI in EM economies constant.
To check whether this approach is robust using out-of-sample data, we contrast the actual path of EM inflation with the scenarios developed in April 2011. We see that over the last year EM headline inflation has hovered between our moderate and our adverse scenario (see Exhibit 5). This confirms our ex ante assumption that food inflation would remain the most important determinant of EM headline inflation, and also provides a level of comfort that our estimation approach and results are fairly sensible. It broadly confirmed our estimates for a lag of about four months in international food prices feeding through to EM inflation rates on aggregate.
EM inflation set to increase more moderately than in 2010-11
With our two scenarios for international food prices, and our updated pass-through coefficient, we can now calculate two potential paths for EM food inflation. Using these, we then turn to estimating the impact of EM food inflation to EM headline inflation. To do this, we use the relevant food weights to split EM headline inflation into a food and an ex-food component. We then assume that EM inflation ex-food continues to grow at the current pace and we add the weighted path of food inflation to project the headline rate. We find:
- Relative to the latest available inflation data (August), there may be further downside to aggregate EM headline inflation due to food contributions. The impact of base effects and the relevant lags between international and local food prices imply that we may need to wait until the full set of October inflation prints are out to fully confirm the beginning of the systematic pick-up in EM food inflation.
- From October onwards inflation starts to rise and peaks, on a year-over-year basis, in March 2013, i.e., 40-60bp above current levels and 80bp-100bp above the projected trough. After March 2013, inflation starts to decline. The pace of the decline will depend on future weather conditions. A moderate weather environment would lead to a quicker and deeper normalisation in EM inflation.
- Our projections suggest the peak in headline inflation will be lower than the 2011 food price spike episode, at between 4.6% and 4.8%yoy depending on weather conditions, compared with 5.1% in mid 2011. This is mostly because the food price increase itself is projected to be somewhat smaller for international food prices on aggregate and in annual terms, and to be less broad-based (focused on wheat, corn and soy). In addition, non-food inflation rates in the first half of 2011, when EM headline inflation peaked, were slightly higher (about 20bp on average) relative to the current annual pace of non-food inflation.
There are three key risks around these conclusions.
- Timing appears to be more uncertain this time around. As mentioned earlier, there are signs across a number of EMs that food inflation is already picking up. This may mean that the lag estimate of four months in the pass-through from international to local food prices may be too lengthy this time around. In turn, this means that EM food inflation is likely to pick up sooner than October.
- Relative to the last food price spike in 2011, this analysis may be less applicable to Asian economies. This is chiefly because of the much more stable price developments in rice. To some extent our analysis takes this into account; as mentioned earlier, we map the corresponding shifts in the corn, wheat and soy basket on broader shifts in the S&P GSCI® Agricultural Index. And this is, in part, the reason why the size of the shock in aggregate international prices is smaller. However, we are conscious that we run our exercise on a high level of aggregation, which does not allow for more precise adjustments along those lines.
- The uncertainty in non-food inflation may be high in the months ahead. Oil prices are expected to recover from current lows but a lot will depend on the pace of global demand and developments in geopolitical risks. Moreover, there is a degree of co-movement between food inflation and core inflation across several EMs, which may pose upside risks to our stable current non-food inflation assumption. Finally, core inflation may exhibit a high degree of variation across emerging markets. We are coming out of a period of softening growth in EM economies which could dampen headline inflation prospects. That said, many EM economies continue to run at high rates of capacity utilisation and experience persistent inflation inertia.
Note that these assessments do not constitute an inflation forecasting exercise but rather an illustration of likely paths for food-driven EM inflation on aggregate. There are, of course, local particularities that may create deviations from such assessments on a regional or country level. Our Asia and CEEMEA Economics research team have also done quantitative work projecting the likely impact of higher food prices on local CPIs. Reassuringly, their findings are broadly consistent with ours; in CEEMEA, our economists expect a 50bp-100bp upside contribution to headline inflation, mostly due to higher food prices but also accounting for the impact of energy prices. In Asia, our economists expect food inflation to add 100bp to local inflation.
EM currencies to benefit
Given the significance of food inflation for overall headline inflation levels and the linkages between food and non-food inflation recorded in the past, EM central banks are unlikely to fully dismiss food price volatility as a temporary and mean reverting phenomenon. Instead, they are likely to respond by tightening monetary conditions either via guidance (a more hawkish stance) or via currency strength (to curtail price pressures on imported food items), or even via higher policy rates. As international food prices are available in high frequency, markets are likely to anticipate these shifts to some extent. Given, however, that ex ante market assessments are conditioned on a number of underlying macro developments, shifts are likely to be priced only partially.
Therefore, it is reasonable to expect market shifts to occur as EM food inflation pushes headline inflation up and EM policy makers react proportionally. Overall, higher headline inflation in EMs is broadly consistent with higher front-end rates (or rate expectations), flatter EM curves and currency strength. To confirm this intuition, we run a simple cross-asset event study of the last three food inflation spikes: 2004, 2007-08 and 2010-11 (Exhibit 7). We examine the average impact of food-driven headline inflation on EM curves and currencies, and also look at equity market behaviour.
More specifically, to proxy for shifts in near-term interest rate expectations, we look at the change in 1-year rates 1-year forward relative to the US (to account for global shifts in fixed income markets). We also look at shifts in the spread between 5-year and 2-year EM rates relative to the US to proxy for shifts in the broader shape of the curve. Lastly, we examine average EM FX returns vs the USD and average EM equity performance vs the SPX. Arguably, it is hard to rely on such small sample assessments and cross-EM averages, but it is interesting that our results generally confirm our macro intuition:
- Typically, 1-year 1-year forwards tend to increase on average, albeit by a small amount, while EM curves flatten significantly in only two of the three episodes.
- EM currencies appreciated strongly vis à vis the USD during the last two food inflation spike episodes and were flat in the first episode under study.
- Interestingly, EM equities outperformed the SPX in all three episodes. It is hard to argue that such a negative supply shock can be linked to benign equity market trends. Indeed, in absolute terms, equities fell in two of the three spikes. The relative outperformance may be due to stronger EM growth vs G10 in our sample.
Hard as it may be to draw firm conclusions from a limited sample, EM FX vs USD strength appears to be the clearer tradable result of EM food inflation pressures. Forward rate expectations have also tended to pick up, albeit to a small extent, while curve flattening is less obvious. Lastly, it is not clear if we will observe a repeat of the relative EM equity strength we saw in the past given the current mixed cyclical backdrop across different EMs.