CasP Dialogue 2015.01: Baines' work on food price inflation

Post your research in political economy, share ideas and workshop your empirical work.

Moderator: sanha926

Re: CasP Dialogue 2015.01: Baines' work on food price inflat

Postby josephbaines1 » Wed Dec 09, 2015 2:47 pm

Hi James and everybody else,

Yes, it has been a fantastic response. I'm really honored to receive these rigorous and thoughtful comments. However, I'm not overwhelmed, so keep the contributions coming! This is the last week of Autumn term at the LSE, and because I've got quite a lot of teaching to do in the next few days I won't be able to formulate a response until early next week. Hopefully my reply will bring some kind of order to the different observations that the contributors have made so far and discrete themes will emerge. But for now, I suggest that we just keep this thread open to all streams of thought; we can create sub-threads later if need be.

All the best,

Joseph
josephbaines1
 
Posts: 16
Joined: Mon Nov 25, 2013 3:35 pm

Why CasP?

Postby Jonathan Nitzan » Wed Dec 09, 2015 4:31 pm

Why CasP?

1. What is the advantage of a CasP approach to this subject?

2. Will (can) a neoclassical analyst or a Marxist political economist (classical or neo-) ask the same questions?

3. Will (can) they derive the same conclusions?

4. What will be the overlaps/differences?

5. Do they matter?
Jonathan Nitzan
 
Posts: 38
Joined: Sat Dec 06, 2008 2:39 pm

Re: CasP Dialogue 2015.01: Baines' work on food price inflat

Postby blairfix » Sun Dec 13, 2015 8:42 pm

Baines' work highlights the important role of biofuels in maintaining the profitability of agribusiness. But biofuels also likely bolster the profitability of big oil. Why? For the simple reason that biofuels are not a net source of energy. Every joule of energy obtained from corn ethanol requires about a joule of fossil fuel inputs.

Mario Giampietro and Kozo Mayumi have written an excellent book on the energetics of biofuels, called The Biofuel Delusion. While showing that the goal of large scale biofuel production is absurd they do not analyse why it is being implemented. Baines' work has much to contribute on this topic.
blairfix
 
Posts: 88
Joined: Fri Dec 20, 2013 3:45 pm

Re: CasP Dialogue 2015.01: Baines' work on food price inflat

Postby jmc » Sun Dec 13, 2015 10:12 pm

Like some of the other posts in this thread, mine will explore research questions that Baines’ paper suggests. To be clear right at the beginning: this is not a critique of Baines’ paper, as there is always something that one’s research does not cover. Rather, I think that Baines helps us avoid falling into platitudes about the nature of capital accumulation in the food business.

Baines convincingly challenges two facets of the “Supermarket Thesis” (ST), one theoretical and the other empirical. At the theoretical level, ST touches on the power of concentrated ownership, but ultimately assumes that advantages in purchasing and pricing are disturbances to an otherwise “normal” economy. Baines’ summary of the capital-as-power approach need not be repeated here, but it, in my opinion, effectively argues that CasP can re-frame the role of power in an analysis of food price inflation. At the empirical level, Baines’ paper demonstrates that ST was left on the shelf past its expiry date (nailed it!). ST has little relevancy in the twenty-first century, and for the researchers that overlook this change in importance, they will be unable to account for the effects of the Agro-Trader nexus on food price inflation and world hunger.

Yet there are still some questions about the relevancy of ST. I have no horse in this race, but I have a question of curiosity:

What about the other peaks and troughs of food prices in the twentieth century?

As Baines makes clear, part of the problem is that ST was (becoming) popular in the late 1990s and 2000s, just around the time when, thanks to Baines’ historical analysis, the food sector experienced shifts in intra-capitalist struggle. However, it is less clear to what degree that ST has lost empirical relevancy, especially because the historical narratives state that the larger supermarket chains originated in the mid-1960s, which is just before the largest wave of inflation in Figure 1 (from ~40 to ~120). Who were the winners and losers of that wave?

While the theory of ST mastery might be too general to be helpful, Baines does have a general claim of his own:
More generally, one can see that in the whole period of severe food price inflation (shaded in grey) the Agro-Core and TraderCore differentially accumulate much more rapidly and for a much longer period of time. Indeed, as shown by the statistics presented in Figure 4, from 2002 to 2012 the major food retailers and manufacturers have only been increasing their differential profits relative to dominant capital by a paltry two per cent a year; while the Agro-Core and the Trader-Core have increased their differential profits annually by an astounding 20 per cent and 27 per cent, respectively. These observations suggest that in periods of rapid agricultural commodity price inflation, the firms that are in closer proximity to the end-consumers in food supply chains find it hardest to differentially accumulate (my italics, page 14).
As I interpret it, the “periods of rapid agricultural commodity price inflation” refer to the peaks in the Economist Food Index—but is it generally true that, in periods other than the most recent wave of inflation, firms “in closer proximity to the end-consumers in food supply chains find it hardest to differentially accumulate”?

My question goes slightly outside Baines’ scale of analysis, but it is because my re-read of his article fed my curiosity. Moreover, I can imagine that it can be a challenge to speak at a more granular level of history when some data on food prices and indices will cover centuries. So any thoughts or retorts are welcome.

P.S.,
As an aside, I went looking the Economist data myself, but I think I need to be an active subscriber to download the data set. However, I did find data on wheat prices in the UK, which have similar waves of inflation in the twentieth century. Below is a figure of wheat prices in the UK, weighted by the UK CPI. To better understand the intensity of price inflation, the series is presented as an annual rate of change from a 20-year trailing average. If the agricultural and trading sectors of the food-business chain benefit from rising wheat prices, the peak in the 1960s suggests ST is, once again, not exactly the story that needs to be told.

Image
jmc
 
Posts: 40
Joined: Mon Oct 19, 2009 2:33 pm

Re: CasP Dialogue 2015.01: Baines' work on food price inflat

Postby PhilHoward » Mon Dec 14, 2015 5:04 pm

I have studied these industries for nearly two decades, and I first read this article almost two years ago. It introduced me to Capital as Power, and encouraged me to read much more. I became so interested that I employed a CasP approach in my forthcoming book on economic concentration in the food system.

This article is an excellent synthesis and critique of the agri-food literature, and I have recommended it to many of my colleagues working in this area. I agree that the quantitative methods Baines uses to explore power shifts are an important advance over previous studies. The evidence for food price inflation as redistribution is extremely important. It suggests researchers should devote much more attention to alliances between firms, and the restrictions they cooperatively place (in other words, the sabotage they inflict) on almost everyone else.

My critique, however, is that I don’t think the characterization of supermarket power in previous studies was nuanced enough. The citation for Burch and Lawrence’s acceptance of supermarkets as masters of the food system (2009: 268) actually states:

“Moreover, our acceptance of the view that supermarkets were the ‘new masters of the food system’ (Winson 1993) and the main locus of control in terms of the establishment and management of supply chains, tended to diminish the importance of new actors who were challenging the dominant role of the retail sector.”

Burch & Lawrence restated this more critical view of “mastery” in 2013 (p. 247):

“In an earlier paper (Burch and Lawrence 2005) we argued that power in supply chains had shifted from the food manufacturers to the supermarkets and the retail sector, creating significant impacts on farmers, processors and consumers. However, we raised a number of questions about the nature of this transformation, and suggested that this did not necessarily involve a permanent shift in control. We argued that while the supermarkets could, in Winson’s (1992) terms, be characterized as ‘the Masters of the Food System’, there were other players emerging who were in a position to challenge that dominant role.”

I am a bit wary of overemphasizing a decline in food retailer power. The relatively short recent period in which the sector has experienced stagnation does not change the fact that these firms still exert a tremendous amount of control over very large segments of society (as Baines notes in the conclusion), even if their relationships with large manufacturers are stabilizing (keep in mind that many food processing/manufacturing industries consolidated in the last few decades with a publicly stated rationale of gaining leverage with Walmart). In addition, in the U.S. context, which I follow more closely, Walmart has stagnated, but the second-largest firm, Kroger, is rapidly increasing its power. As Baines notes in his other works, Kroger has adopted many of the strategies implemented by Walmart, which has contributed to the former more than doubling its market cap in the last few years.

I don’t have access to Compustat data, but have observed declining market caps of many firms in the Agro-Trader Nexus (e.g. Monsanto, ADM, Deere & Co.) since the article was written – I therefore suspect that updating the analysis would find rates of accumulation that are moving closer to those of the retailing and processing sectors.

Perhaps another interesting question to follow up on is if any of these food firms are approaching the asymptotes of power described by Bichler and Nitzan. Baines described the social resistance to genetic engineering, which was partially overcome by seed/chemical firms. Fewer genetically engineered seeds have been commercialized in recent years, however, and sales are declining. As a result, there has been a tremendous amount of jockeying to consolidate the seed/chemical industry as an external path to growth (Syngenta managed to rebuff numerous buyout offers from Monsanto and China National Chemical Corp., but DuPont and Dow just announced plans to merge and then break up into three more specialized firms – each with a more dominant market share in their respective industries). The ETC Group (2015) suggests that a “feeding frenzy” of acquisitions is about to begin, possibly including farm equipment firms buying up seed/chemical firms to become “one-stop shops” for inputs.

Will these drastic measures be enough to prevent food industries from returning to “a dead skunk at a tea party” (p. 97) in the eyes of investors? Even if they are not, how much damage will they inflict upon society in the process?

This article persuasively documents a number of harmful impacts that are increasing, particularly the institutionalized waste of biofuels and its role in hunger. Raising public awareness of these issues, as well as the role of dominant food firms in many other social and ecological problems is a key task, so I hope these discussions can enroll many new participants.

Phil Howard
https://www.msu.edu/~howardp/

References:
Burch, D., & Lawrence, G. (2013). Financialization in agri-food supply chains: private equity and the transformation of the retail sector. Agriculture and Human Values, 30(2), 247-258.

ETC Group. 2015. Mega-mergers in the global agricultural inputs sector: threats to food security & climate resilience. http://www.etcgroup.org/content/mega-me ... uts-sector
PhilHoward
 
Posts: 5
Joined: Thu Dec 03, 2015 2:49 pm

Re: CasP Dialogue 2015.01: Baines' work on food price inflat

Postby DT Cochrane » Sun Dec 20, 2015 8:44 pm

Some great - and provocative - responses so far.

Catalyzed by a line from Sandy's post, I want to consider the boundaries of a 'corporate coalition' and how we define that boundary.

We can think about the qualities of accumulation in differential terms just as we do the quantities of accumulation. In fact, differential accumulation necessitates it. I suggest that one of the most important analytical roles of political economy ought to be identifying the relevant differences and demarcations between co-operators and competitors. Joe has articulated some of the qualities among the four -cores that compete within the food sector. However, the question emerges, how do we articulate the core within a segment? Compared to the detail and precision of Joe’s articulations among the groups, the demarcation of ten firms (for three of the -cores) feels arbitrary. It is understandable why Joe did it for practical purposes, but perhaps we need to devise new measures and methods to delineate the dominant members within a qualitatively differentiated segment.

As a brief aside, Shai, this might be a suitable - and important - task for the technical/mathematical skills that you have to share. Are there methods already existing, or capable of being devised, to distinguish the inside and the outside of dominance within a cohort?

My thought on this matter came from Sandy's innocuous - and entirely accurate - observation that "Baines shows that supermarkets ... have seen their relative share of profits decline since 2000." My immediate thought was, "Well, differential not share...." Then, I realized these were synonymous owing to the structure of Joseph’s cohorts. Differential capitalization is calculated as a fraction of fractions. The numerator has the per firm capitalization of the segment under analysis. The denominator as the per firm capitalization of the chosen benchmark. You can re-arrange this to create two new terms. The first is the segment’s share of the benchmark’s capitalization[1], the second is the inverse of the segment’s share of the total number of firms. So, differential capitalization increases if the segment’s share of capitalization increases OR its share of the number of firms decreases. However, because the number of firms in both the segment Joseph constructed and the benchmark are unchanged, that second term is unchanged. But, what if the -core has actually undergone concentration? Is 10 firms still the appropriate number for the calculation? If Joe had used a different differentiator - one that changed the number of firms within the -core - then the differential values may have changed, potentially in significant and meaningful ways.

[1] This assumes that all the members of the segment are also included in the benchmark.
User avatar
DT Cochrane
 
Posts: 66
Joined: Thu Nov 21, 2013 6:49 am

Re: CasP Dialogue 2015.01: Baines' work on food price inflat

Postby shai » Mon Dec 21, 2015 12:49 am

Per Troy's aside:

Indeed there are methods that deal with similar questions. Two preliminary (interconnected) answers I can think of are:

1. In 'network science' there are common measures of, e.g., 'centrality' or 'community structure' (for a brief introduction to networks see http://www.capitalaspower.com/2014/08/cs-pe-networks/)

2. 'Cluster analysis' machine learning algorithms may be useful (where clustering may be 'fuzzy', allowing objects to be in more than one class).

The rub lies less in computing the quantitative measures, but rather in the preliminary definitions. For networks, one has to define 'vertices' and 'edges' in a meaningful way. Cluster analyses are generally based on some quantitative notion of 'distance'.

A more thorough discussion on possible such definitions may be a candidate for a thread later on.
shai
 
Posts: 5
Joined: Mon Oct 18, 2010 2:48 am

Re: CasP Dialogue 2015.01: Baines' work on food price inflat

Postby josephbaines1 » Mon Dec 21, 2015 12:20 pm

In this entry, I intend to prod away at a key pillar of the argument put forward in my paper ‘Food Price Inflation as Redistribution’ (FPIAR): that rising agricultural commodity prices of the 2000s contributed to the relative stagnation in the profits of major supermarkets. My renewed interest in this contention is prompted by the insightful comments of the contributors to the dialogue. The entry points our attention to the fact that when we analyse the distributional dynamics of relative price changes, we should examine not only differential profits, but also what Nitzan and Bichler call differential ‘depth’ and differential ‘breadth’.

Breadth is registered in firms’ employee numbers and depth is defined as profits per employee. As Figure 1 shows, breadth and depth are the two major axes of corporate earnings. Firms can increase their earnings by expanding their employee base through mergers and acquisitions and green-field investment; and they can increase their profits through cutting costs of inputs/inventories or through increasing costs of outputs/finished products (which in turn affect sales and expenses per employee). Partly due to space constraints in my original article, I did not consider the profit trajectory of major food retailers with reference to these two key axes of accumulation. The issues of breadth and depth are perceptively alluded to by both Cochrane and Gorsky.

F1.png
F1.png (254.57 KiB) Viewed 1599 times


By examining the breadth and depth dimensions of food retailers’ relative profits, we can establish whether the relative stagnation of supermarkets and wholesalers that I identify in my original article was brought about principally by the slowdown in their breadth expansion, or principally by a diminution in their depth, i.e. their capacity to cut costs or to pass on price rises to consumers in the inflationary conditions of the 2000s. If it is the former, then my original argument is weakened; if it is the latter, then my original argument is strengthened; and if both play an important role, then more work needs to be done in analysing the relationship between breadth and depth dimensions of supermarket profitability, and therefore the original argument needs to be refined.


Supermarket Breadth and Depth Relative to the Agro-Food Sector

Figure 2 represents an initial attempt to measure the relative depth of major supermarkets. Following Figures 2 and 3 of FPIAR, I draw on data from Datastream. This source has multiple strengths. It provides much more historical data than Mergent or Bloomberg, much more wide-ranging data than CRSP, and much more complete data series than Compustat. Datastream has its own equity index for ‘food wholesalers and retailers’. As of the beginning of 2015, this index comprised 80 firms. Interestingly Wal-Mart is not one of them. The absence of Wal-Mart in the index is important because, as Cochrane and Howard astutely point out, the ‘Retail-Core’ proxy that I constructed for Figure 4 in FPIAR is dominated by Wal-Mart, and as a result its accumulation trajectory closely follows the accumulation trajectory of this one mammoth firm. In the ‘food wholesalers and retailers’ index, we have a proxy for major supermarkets (and wholesalers) that is not skewed by Wal-Mart’s idiosyncratic pecuniary gyrations.

F2.png
F2.png (611.29 KiB) Viewed 1590 times


As you can see from the chart, I adopt Nitzan and Bichler’s measure for differential depth: relative profits per employee - calculated by dividing major food retailers’ and wholesalers’ weighted average profit/employee ratio by the weighted average profit/employee ratio of all firms in the agri-food sector listed in Datrastream. I also plot the supermarkets’ relative profit margins (relative profit/sales ratios), calculated mutatis mutandis.

Two major observations can be made from the figure. First, supermarkets’ differential profit margins are closely correlated with their differential depth. The strong correlation (r=0.89) is attributable to the fact that differential profit margins represent a key component of differential depth (see Nitzan and Bichler 2002, 171-2). Second, the trajectory of the food retailers’ and wholesalers’ differential profit margins and differential depth broadly follow the narrative arc of the first and second sections of FPIAR, in which retailer power is said to reach a zenith in the agri-food system at the turn of the twenty-first century and decline thereafter.

But what about the breadth dimension of supermarkets’ profits? Figure 3 traces food retailers’ and wholesalers’ share of workers employed in the agri-food sector. It also tracks their share of sectoral revenue. The chart reveals explosive growth in relative employee and sales levels in the last five years of the twentieth century. At the beginning of the twenty-first century, food retailers and wholesalers employ almost 65% of people working in listed firms operating in the agri-food sector. The share trends downward thereafter. While there is a steep peak for supermarkets’ employee share in 2001, in supermarkets’ revenue share data we see more of a plateau that inclines slightly upward until the end of 2007, and which then gives way to a precipitous fall in the remainder of the decade.

F3.png
F3.png (350.51 KiB) Viewed 1590 times


The findings made so far have mixed implications for our assessment of the argument advanced in FPIAR. On the one hand, they bolster my original critique of the supermarket mastery thesis insofar as the major food retailers’ relative profits was falling into negative territory in terms of both depth and breadth during the early and mid-2000s - precisely when the idea of supermarket mastery began to have widespread currency. On the other hand, in decomposing food retailer and wholesaler profits in their breadth and depth dimensions, I complicate my original analysis of why supermarkets were experiencing relative stagnation. The original argument, at least implicitly, suggested that the millennial downturn of supermarkets stemmed from problems in the depth dimension, as food retailers’ struggled to cut costs and pass on price increases to consumers in the inflationary conditions of the 2000s. While the data put forward in Figure 2 adds weight to this thesis, clearly it does not tell us the whole story. To recapitulate: Figure 3 shows that since the early 2000s, major supermarkets have also been experiencing a crisis of breadth expansion.

The breadth crisis, as Cochrane kindly pointed out, has been addressed in a project I’ve done specifically in relation to Wal-Mart - a project, I hasten to add, that was kick-started by a post he made in this very forum. But it is worth putting this breadth crisis in context. While major supermarkets’ shares of sales and employees in the agri-food sector have fallen in proportional terms by 3 percent and 7 percent respectively from 2000 to 2015; their differential profit/sales and profit/employee ratios have declined by 42 percent and 40 percent. As such, the relative stagnation of supermarkets has principally been brought about by a differential depth crisis rather than a differential breadth crisis.


Supermarket Breadth and Depth Relative to Global Capital

I now want to turn to an issue that Hager raises - that struggles within the capitalist political economy are multifaceted and that we need to clarify the multiple and ever-shifting lines of these struggles. Cochrane makes a point that compliments this position, arguing that the fluidity of business conflict behoves us to be reflexive about the analytical categories that we adopt. These points direct our attention to the qualitative aspects of research on business power. However, they also have clear implications for the forms of quantitative analysis we engage in, and for the findings to which these quantitative explorations lead. To illustrate, consider Figures 4 and 5. They use the same depth, breadth, profit margin and employee data for food retailers and wholesalers used in Figures 2 and 3. But instead of creating differential measures through dividing these data against the corresponding data for all corporations listed by Datastream in the agro-food sector, it does so against all corporations in all sectors.

F4.png
F4.png (469.46 KiB) Viewed 1590 times


F5.png
F5.png (335.6 KiB) Viewed 1599 times


Although the differential breadth and sales trajectories depicted in Figure 5 broadly concur with those presented in Figure 3, the differential depth and profit margin trajectories depicted in Figure 4 are very different to those revealed in Figure 2. Rather than seeing quadratic lines of best fit that reach apexes at the turn of the twenty-first century, we see curves whose maximal points are reached in the early to mid-1990s. Moreover, the cyclical upturns that punctuate these curves are much more prominent than any upturn presented in Figure 2. In short, by simply changing the benchmark of supermarket performance from the global agro-food sector to global capital at large, we get a different impression of the trajectory of supermarket power. Major supermarkets, from the view of Figure 4, reached the acme of their power over prices in the early 1990s and there has been a secular downturn since. This finding fits in quite well with the data presented in Figure 4 of FPIAR, where the ‘Retail-Core’ is shown to be experiencing little more than relative stagnation for the last two and a half decades.

But what of the sharp cyclical upturns presented in Figure 4 of this entry? They appear to broadly coincide with the onset of economic downturns; specifically, the recession of 1990-91, the bursting of the dot-com bubble in 2000-01, and the Great Recession of 2007-09. Why would supermarkets perform relatively well during these periods? This question demands thoughtful analysis, but it seems at least partly attributable to the fact that consumption of the kinds of basic necessities that supermarkets provide is relatively resilient during periods of economic contraction. In other words, consumer demand for supermarket products may be more inelastic than in other sectors. As such, prices may not have to be adjusted downward significantly by supermarkets to maintain satisfactory levels of business traffic in recessionary conditions. The relatively robust share of sales data for supermarkets during the recessions of 1990-91 and 2000-01, as presented in Figure 5, corroborates this claim.

Again, it is worth comparing how shifts in supermarkets’ relative breadth compares with shifts in their relative depth. Whereas major supermarkets’ shares of sales and employees in listed companies have fallen in proportional terms by 5 percent and 7 percent respectively from 2000 to 2015; their differential profit/sales and profit/employee ratios have declined by an astounding 78 percent and 72 percent. Thus, whether we use the agro-food sector or global capital as the comparator, the general picture remains the same: relative stagnation of supermarkets has principally been brought about by a crisis in food retailers’ differential depth rather than a crisis in their differential breadth. In this respect, the argument put forward in FPIAR seems structurally sound.

With that said, there is perhaps an observation that one can make from Figure 5 that is more troubling. The last major upsurge in the profit margins and depth of supermarkets, relative to global capital as a whole, began in 2007 when food price rises accelerated. This finding is at odds with the overall thrust of my contention that food price inflation was a key contributor to the relative stagnation of supermarkets. Moreover, if we inspect the data in Figure 2 of this post a bit more carefully, we can see that although the agricultural commodity price inflation of the 2000s broadly coincided with a drop in the profit margins and depth of supermarkets relative to the agri-food sector as a whole, in 2007 and 2008 there was an uptick in the profit margins and depth of supermarkets relative to the agri-food sector. However, the relative profit margin and depth data depicted in Figure 2 and Figure 4 show that in the next agricultural commodity price spike – in 2010-11 – no upsurge in the differential depth and profit margins of supermarkets occurred.

As such, these data suggest that a more nuanced conception of the distributional dynamics of food price inflation may have to be developed. This need for nuance is further underscored by the following excerpt of a Bloomberg Business TV interview in 2012 with Michael Schlotman, the Chief Financial Officer of Kroger – a company that Howard mentions in his contribution to this forum:

Schlotman: “At this point we would expect 2013 to be about where we are right now in the mid-one percent range for inflation”

Interviewer: “So in some ways this is almost a sweet spot of a level for you?”

Schlotman: “It’s not too bad. Actually if it was just a touch higher it would be a little bit more of a sweet spot. A little bit of inflation is okay, but none is bad, and too much is bad. What is really bad is when we have some voluble swings like we had a couple of years ago, where it was very high and then went low to negative very quickly. That’s a more difficult environment to operate in.”
(Bloomberg Business, 2012)


In lieu of a Conclusion

This excerpt brings us to the beginning of the end of my initial response to some of the comments made by those involved in the forum. I have only responded to a small portion of the comments so far, and I have raised more questions than answers. These questions include the following: is the relative stagnation of major supermarkets witnessed in recent years a secular phenomenon brought about by some asymptote being reached in the differential breadth of food retailing? Or is the relative stagnation more of a cyclical effect of the upsurge in agricultural commodity prices in the 2000s? Can the neoclassical concept of demand elasticity invoked in this post be recast in the terms of business sabotage and capital as power? What are the transmission mechanisms between price rises in raw agricultural commodities and the prices paid by consumers for products bought at the supermarket checkout? Should price volatility, rather than the level of prices, be the primary consideration when we analyse the distributional effects of price changes? And finally, what is the relationship between the differential breadth of supermarkets and their differential depth?

In considering the relationship between breadth and depth, we’d do well to recall the presentation that Joe Francis gave in a conference organized by the Forum on Capital as Power back in 2011. In this presentation, Francis sketches a theory which links breadth regimes of accumulation with increased downward pressure exerted by oligopsonic firms on suppliers’ prices throughout the commodity chain. Their breadth expansion is emulated by the breadth expansion of firms operating upstream in supply chains as a response to increased oligopsonic power (a point that Howard makes in his contribution, specifically in relation to Wal-Mart). Due to the limitations on mergers placed by anti-trust law, the decline in attractive takeover targets, paroxysms of economic nationalism, and other developments, the breadth regime may reach a point of exhaustion. In this situation, equity markets become less attractive for investors, and commodity markets become subject to a speculative frenzy that helps to drive up commodity prices. These commodity price hikes necessitate the breadth expansion of oligopsonic buyers so that they can regain some control over input prices; and the cycle begins anew.

In the next post, I plan to test some of these claims. Specifically, I’ll ‘zoom in’ on the distributional effects of price changes through revisiting Figure 3 of FPIAR, and through producing a more granular analysis of differential prices and differential profitability. And I’ll ‘zoom out’ by contextualizing the distributional dynamics of food price changes in a broader disaggregate analysis of global breadth and depth regimes. These moves will necessitate adopting a wider historical lens – a move that McMahon convincingly proposes in his contribution. Moreover, in adopting a wider historical lens, I’ll have to use the patchier but nonetheless longer time-series data provided by Compustat. As for comments regarding the Agro-Trader Nexus and the biofuels boom, I’ll get there eventually. For now, it suffices to say that this charted survey shows that one of the key pillars of the argument in FPIAR regarding the linkage between supermarket stagnation and food price inflation may have somewhat thinner evidentiary foundations than previously thought. The pillar still stands, but it requires refinement and support from more research.


References:

Bloomberg Business. 2012. Interview with Michael Schlotman, Chief Financial Officer of Kroger. 7 March 2012. Available from: http://www.bloomberg.com/news/videos/b/ ... c9d6a48d4e (accessed on 21 December 2015).

Nitzan, J. and Bichler, S. 2002. The Global Political Economy of Israel. London: Pluto Press.
josephbaines1
 
Posts: 16
Joined: Mon Nov 25, 2013 3:35 pm

Re: CasP Dialogue 2015.01: Baines' work on food price inflat

Postby shai » Sun Jan 10, 2016 8:09 pm

Some notes on cluster analysis and dominant capital

Following some of the points that were raised in the thread, Joseph and I have in the past couple of weeks been exploring the prospects of applying 'cluster analysis' methods hoping to disaggregate data on large groups of firms into 'dominant' and 'non-dominant' firms.

In a nutshell, cluster analysis algorithms attempt at identifying sub-groups inside a group of objects based on some notion of affinity. In practice, main notions of affinity are distance, density and likelihood of resulting from specific probability distributions.

Figure 1 demonstrates how different clustering algorithms classify different datasets. The first three rows offer a comparison over different manifestations of 'clusters' while the last row examines how the algorithms perform on random, non-clustered data. The algorithms differ not only by the results they provide, but by the assumptions that underlie them and the choice of parameters.

Figure 1
cluster_comparison.png
cluster_comparison.png (821.28 KiB) Viewed 1400 times


source: http://scikit-learn.org/stable/modules/clustering.html

To use some of the algorithms, one must provide the number of clusters the algorithm should divide the data to, while it is built into some other algorithms to choose the 'optimal' number of clusters when running. Some algorithms perform well for a small number of clusters and badly for a large number of clusters, some perform well on small datasets but become impractical on large datasets, and the list of differences that suggest which algorithm is most suitable for which circumstances continues.

For our initial study, then, we chose the BIRCH algorithm (rightmost column) simply for its ease of implementation. Still, with all the subtleties of choosing the correct algorithm, it is manifest from Figure 1 that a successful implementation of any of the above cluster analysis algorithms requires the dataset chosen to indeed be made of separate clusters.

What is the case for COMPUSTAT firms?

The answer is not immediate. The first step is to choose meaningful and significant variables, and no a-priori answer is provided for this question. For the sake of our initial analysis we have chosen four variables per firm: assets, employees, net income and revenues (Figure 1 demonstrates clusters in two-dimensions, for two variables, but almost all algorithms are capable of performing the analysis on any finite number of dimensions/variables).

In this post, we show the results we have obtained for an analysis of the food-manufacturing firms. Figure 2 shows a 3-dimensional scatter plot of three of the variables: assets, employees and revenues for the year 1990 (which demonstrates patterns that are rather representative of most years).

Figure 2
3d-no-cluster.png
3d-no-cluster.png (30.22 KiB) Viewed 1400 times


We see that most of the firms are 'small' and there are some that are 'exceedingly' large. On the face of it, this may seem promising for the sake of identifying 'dominant capitals' as the 'exceedingly large' firms. This indeed may be the case. But clustering algorithms perhaps may not be the best way to perform this analysis. Examining Figure 2, it is rather obvious that there are simply no clusters in any of the senses that clustering algorithms seem to be most able of handling.

Nonetheless, Figure 3 presents the results from the BIRCH cluster analysis with the aforementioned four variables (of whom only three are presented in the figure). The blue points represent one cluster, and the red points the second cluster. True, this distance based clustering algorithm identifies the largest firms as a single cluster. However, given the nature of the data here and considering the tasks clustering algorithms are best provided to handle, we find this analysis to be poorly informative.

Figure 3
3d-cluster.png
3d-cluster.png (30.22 KiB) Viewed 1400 times


To sum: cluster analysis is fun! But at least for the choice of variables presented here (and, actually, any subset of 2 or 3 variables out of the four, not presented here) cluster analysis seems not to be the appropriate means of analysis. Perhaps a different choice of variables may be more suitable, and perhaps a different mode of analysis will be more useful for these variables.
shai
 
Posts: 5
Joined: Mon Oct 18, 2010 2:48 am

Re: CasP Dialogue 2015.01: Baines' work on food price inflat

Postby josephbaines1 » Tue Jan 19, 2016 4:45 pm

In my first post in this dialogue, I re-examined the arguments put forward in my paper ‘Food Price Inflation as Redistribution’ (FPIAR) regarding the distributional effects of food price rises in the 2000s. In this post, I narrow the focus and address an issue that James astutely raises in his post. As James points out, in FPIAR I suggest “that in periods of rapid agricultural commodity price inflation, the firms that are in closer proximity to the end-consumers in food supply chains find it hardest to differentially accumulate”.

The claim was partly based on my findings regarding relative income shifts in the 2000s. And it was also based on my view, never explicitly articulated in FPIAR, that a rising agricultural commodity price environment would enhance margins for those upstream in food supply chains, at the expense of those operating downstream. Why? Because end-consumers would be less sensitive to firms operating upstream putting on additional margin on top of rising agricultural prices. However, for firms operating in greater proximity to the end-consumer, a rising agricultural commodity price environment would present itself less as an opportunity to tag on additional margins, and more as a cost that they would have to at least partly absorb in order to maintain satisfactory levels of business traffic. James rightly challenges me to look at “the other peaks and troughs of food prices in the twentieth century”, to ascertain whether this logic stands up to scrutiny. Let us call the general claim of mine ‘the redistribution-inflation hypothesis’ and let us now reckon with James’s challenge.


The distributional dynamics of the 1970s and 2000s agricultural commodity booms compared

Figure 1 traces the shifting relative profit margins of major supermarkets, major food manufacturers, major agro-chemical companies, and major agribusiness producers. It should be noted that these firm subsets are different to the retail-core, food-core, agro-core and trader-core proxies I use in FPIAR in three respects. Firstly, while the retail-core and the food-core represent the ten highest earning food retail and food manufacturing firms respectively, the ‘major food retailers’ and ‘major food manufacturers’ designations are represented by the five highest earning firms of each respective cluster. Secondly, while the agro-core proxy comprised a broad constellation of the five highest earning agricultural input firms, including pesticide companies, agro-biotechnology corporations and farm machinery manufacturers, the ‘major agro-chemical firms’ subset comprises a more specific constellation of the five highest earning firms that produce agricultural chemicals. Thirdly, while the original trader-core proxy comprised a particular constellation of the three most profitable agricultural commodity traders, the ‘major agribusiness producer firms’ is a designation that encompasses the five most profitable firms directly involved in agricultural production. The original trader-core proxy has been discarded because of the lack of available data prior to 1996.

c.png
c.png (535.73 KiB) Viewed 1212 times


c6.PNG
c6.PNG (25.89 KiB) Viewed 1226 times


The above table lists the constituent firms in each category for 1974 and 2014. A number of comments can be made here. You can see that Monsanto - a major agrochemical firm - is put in the ‘Agricultural Production’ category (renamed by me as ‘major agribusiness producer firms’) of the Standard Industry Classification because it controls the production of seed. This categorization of Monsanto underscores the fact that classification oftentimes dissimulates the wide-ranging scope of large firms’ operations. Moreover, you can see that ADM is listed in the ‘major food manufacturers’ category. The original food-core proxy was shorn of both of the major publically-listed agricultural commodity trading firms (ADM and Bunge) when it was constructed.

More generally, it is work asking why these new proxies comprise five firms, rather than say seven or thirteen? This issue speaks to a problem that Troy appositely raises in regard to the original proxies I constructed for FPIAR:

“[T]he question emerges, how do we articulate the core [of dominant firms] within a segment [of the agri-food sector]? Compared to the detail and precision of Joe’s articulations among the groups, the demarcation of ten firms (for three of the -cores) feels arbitrary. It is understandable why Joe did it for practical purposes, but perhaps we need to devise new measures and methods to delineate the dominant members within a qualitatively differentiated segment.”

In line with Troy’s subsequent recommendation to develop a less arbitrary method for delineating dominant firms, Shai and I have sought to use two-step cluster analysis to establish new proxies for dominant firms within the different segments of the agri-food sector and to establish a new proxy for dominant capital at large. Our trials, tribulations and discontents with this endeavour were outlined by Shai in his last post. The challenges of constructing a proxy for dominant capital are also discussed by Bichler and Nitzan (2012: 51-52). For the statistically inclined, determining the shifting boundaries of dominant capital, and the shifting boundaries of clusters of firms therein, would perhaps be a very useful challenge to take up in a MA thesis or indeed in a PhD dissertation. Yet for the time being, indexical analysis, rather than cluster analysis, of dominant firms will have to suffice.

With this background information in mind, we can turn our attention to the findings presented in Figure 1. The figure illuminates the distributional shifts between the major firms of the different segments of the agri-food sector more clearly than Figure 4 of FPIAR - in which the pecuniary trajectories of clusters of dominant agri-food capital are first presented. Indeed, in Figure 4 of FPIAR the benchmark is the Compustat 500 - the five hundred largest US-listed firms, ranked by net income. But in Figure 1 of this post the benchmark is the ‘Agri-Food Capital 20’. This benchmark comprises a composite (with weighted parameters) of the 5 largest food retailers, the 5 largest food manufacturers, the 5 largest agribusiness producers, and the 5 largest agro-chemical firms, all ranked again by net income. As such, we see here in Figure 1, the performances of the different firm clusters relative to one another, rather than relative to dominant capital at large.

At one level, the redistribution-inflation hypothesis is corroborated by Figure 1. Indeed, the firms furthest upstream in agri-food supply chains - the major agro-chemical and agribusiness producer firms - register significant increases in their relative profit margins, during the agricultural price boom of the 2000s, and also during the agricultural price boom in the first half of the 1970s. However, at another level, the redistribution-inflation hypothesis is complicated by the data presented in the figures. To be sure, if the redistribution-inflation hypothesis was unimpeachable, the relative margins of the major agri-chemical and agribusiness producer firms would have subsided significantly from 2012 onward in line with sliding agricultural prices. Yet no significant fall in relative margins is evident so far. Such a finding perhaps betokens a long-lasting seismic shift, rather than just a cyclical swing, of power towards firms upstream in agri-food supply chains.

In any case, the markup trajectory of supermarkets relative to food manufacturers also raises questions. According to the logic of the redistribution-inflation hypothesis, the profit margins of the major food retailers would be squeezed to a greater extent than the profit margins of major food manufacturers, due to the greater proximity of the former to consumers. But this hypothesis does not appear to be affirmed by the facts. Although major food retailers experience a general downtrend in their relative profit margins during the agricultural commodity price booms of the early to late 1970s and the early to late 2000s, so do the major food manufactures, and possibly to a greater extent.


Supermarkets versus food manufacturers: a closer look

As we move to the second half of this long overdue post, I’d like to draw the patient readers’ attention to Figure 2. The triumvirate of charts in the figure offers a more comprehensive view of the performance of major supermarkets vis-à-vis food manufacturers. The top chart tracks the profit margins and depth of major food retailers relative to those of major food manufactures; the middle chart tracks the corresponding sales and breadth differentials; and the bottom chart tracks the retailer-manufacturer labour productivity differential. I have opted to include this last category because if we express depth in the terms of profit margins, the residual category is sales per employee; and sales per employee is a ratio used by many (e.g. Wrigley 1992) to determine companies’ labour productivity.

b.png
b.png (536.38 KiB) Viewed 1212 times


However, we should perhaps be circumspect about this proxy for labour productivity because it does not control for variances in unit costs and in labour time. In an ideal world, I would use full time equivalent employment data, and I would use output data that is stripped of the vicissitudes of prices. Whether that can be achieved in disaggregate accounting, or indeed any form of accounting, is a different matter altogether.

These considerations notwithstanding, Figure 2 does provide somewhat enlightening findings. The retailer-manufacturer depth and profit margin differentials were in steep decline in the early post-war period - indicating food manufacturer ascendency over supermarkets; followed by a flat-lining differential in the 1960s - indicating perhaps a retailer-manufacturer ‘truce’ or ‘stalemate’. Then food retailers enjoyed a surge in their profit margins and depth relative to food manufactures during the agricultural commodity price boom of the 1970s - in contravention of what the redistribution-inflation hypothesis would have predicted. With our two indicators then moving sideways, another re-balancing of supermarket and manufactures prevailed in the 1980s. And in the 1990s, we witness the recrudescence of retailer supremacy, as registered in the resurgence in retailer-manufacturer depth and profit margin differentials. Finally, in the 2000s, we see a drop in our indicators, followed by a recovery, especially pronounced for the profit margin differential from 2006 onward. Given that the agricultural commodity price boom lasted from around 2002 to 2012, the pre-2006 drop conforms to the redistribution-inflation hypothesis, but the post-2006 recovery confounds it.

The middle chart, when related to the top chart, also furnishes us with some insights. The retailer-manufacturer breadth and sales differential reach a post-war peak in 1964, a nadir in the early 1990s, and a resurgence that ends at the turn of the twenty-first century. Interestingly, the depth and profit margin differentials appear inversely correlated with the breadth and sales differentials in the1950s and 1960s. However, the negative correlation gives way to a positive correlation in the last two decades. The morphology of this correlation from negative to positive suggests that that the mechanisms of retailer-manufacturer bargaining power may have shifted through time. Finally, the indicator presented in the bottom chart - to the extent that it can be taken seriously as a diagnostic of differential output per employee - suggests that major supermarkets have experienced a chronic decline in labour productivity relative to major food manufactures over the entire period for which we have data. Yet the decline is punctuated by mild uptrends from the early 1960s to the early 1970s, from the early 1980s to the early 2000s, and at the turn of the twentieth-first century.

The task now is to leaven these rather dry quantitative findings with all the insights that a qualitative analysis of corporate power can muster. Specifically, I need to examine the changing historical dynamics of cooperation and conflict between supermarkets and food manufacturers, and I need to analyse how these dynamics refract through the shifting order of differential prices. Furthermore, I have to engage in a comparative analysis of the contested technologies of control by capital over labour in both food manufacturing and food retailing to make sense of the apparent downtrend of relative productivity in the supermarket business.


On the melancholia of research

I’ll end this dispatch with one final remark. Sometimes social-scientific work can cast light on patterns hitherto unseen. But most of the time, all that the researcher encounters are glaring ambiguities. FPIAR did reveal some interesting patterns, but what seemed sharp and well-defined from the distance at which I positioned myself when writing the article, appears fuzzier and more inchoate on closer inspection. A theme thus seems to be slowly emerging: a monocausal analysis of relative profitability in the agri-food system that explains pecuniary trajectories solely in relation to changes in agricultural commodity prices simply will not do. These price changes are but one key aspect of the polyvalences of corporate power.


References:

Bichler, S. and Nitzan, J. 2012. The Asymptotes of Power. Real-World Economics Review. Issue no. 60, pp. 18-53, http://www.paecon.net/PAEReview/issue60 ... tzan60.pdf

Wrigley, N. 1992. Antitrust Regulation and the Restructuring of Grocery Retailing in Britain and the USA. Environment and Planning A. Volume 24, pp. 727-749
josephbaines1
 
Posts: 16
Joined: Mon Nov 25, 2013 3:35 pm

PreviousNext

Return to Research

Who is online

Users browsing this forum: No registered users and 1 guest

cron