This project is building off COMTRADE data to estimate a world trade model for raw and canned tuna. I've split it off from the COMTRADE project to specifically focus on the world trade aspects of the problem.
This piece of work developed from the American Samoa Tuna Policy work. Tuna policy is the Forum Fisheries Authority's domain; however, they are prevented from assisting territories - Guam, American Samoa, Northern Mariana Islands, French Polynesia, New Caledonia, the Pitcairn Islands, and Wallis and Futuna.
The United Nation's COMTRADE system collects all of the world's imports and export information, in local currencies, across all of the System of Harmonised Commodities which goods are traded in. The data is accessible for free using an API and limited per day access.
Import and export data is collected at both the "reporter" and "partner" level. "Reporters" are countries supplying the data, and "partners" are countries on the "flipside" of the transaction. For example, tuna exports from New Caledonia to France would have "New Caledonia" as the reporter, and "France" as the partner.
The currency of the transaction is the "local currency", and to enable comparison between countries, Comtrade expresses all of the transactions in USD equivalent. It does this through data collected in the programme "Comtrade_MetaData.r", which enables both local and international currencies to be derived. From a theoretical perspective, changes in the demand for fish should be reflected in changes in the local currency of the exporter, which is the theoretical source of loss in competitiveness. BUT included in the exporting is a currency effect, which means changes in the currency between local and the USD will also affect the competitiveness. Ideally, we should be able to separate out loss of competitiveness into demand effects (through the local currency), and macro effects (through the exchange rate).
From the importer's perspective, its the combined affect which is relevant to their overall import cost, but from a producers perspective what's eroding their competitiveness is important. In the domestic sphere, tax changes and capacity constraints will lead to price increases which ceteris paribus should feed into international prices, regardless of what else is happening in the domestic or USA economic. However, through the exchange rate, changes in demand for things which dominate the balance of payments will alter the international price outside of fisheries conditions. For example, increases in tourism for a pacific economy might result in an exchange rate appreciation which drives up the price of its fish exports, and (after maybe a lag) leads to a reduction in import demand, irrespective for what's happening in the pacific economy fishing industry.
We should be able to separately identify domestic price effects, and international macroeconomic exchange rate effects.
Its not just a consumer market for fish - raw fish is an input into tinned fish, as well as being a consumer product in its own right. Consequently, there's two products and two connected markets: the raw tuna market, and the tinned tuna market. And good news - we don't have a great idea of the proportion which international each goes into each. From earlier work in the COMTRADE programmes, we know that changes in the two markets are related - world tuna prices for both raw and canned tuna follow similar trajectories. And for countries like Thailand, their raw and canned tuna markets are closely connected than most. But in European markets like Spain, Italy and the Netherlands, these countries are importers and exports of raw and tinned tuna suggesting they have markets where they might be value-adding to both; for example, importing low-price and high-price raw tuna, where low-price goes into canning, and high price goes into something else for on sale.
I don't have a great story, but its a chance to look at these countries in more detail as a complex market for future pacific customers.
I've been scratching for some published models of international trade to base this modelling upon. The paper "Documentation/Regieme Shifting Model.pdf" from "A Two-Regime Model of Exports: U.K. Manufactures, 1980–1996" by Alan King is my initial economic model thinking. In that paper, King notes (as at 2001), the literature splits into 3 essential models of exports demand:
-
Demand Determined, Supply Unconstrained
In these models, traded volumes depend on demand-side variables only, like external competitiveness and foreign economic activity as demand drivers. They assumes exports are not supply constrained. Each country's product is perfectly substitutable between countries. The overall level of demand expressed by any one country depends on only their internal demand conditions. If something like GDP goes up, then export demand goes up.
Which countries receive the demand depends on each country's External Competiveness, with the most price competitive receiving the bulk of the change in demand. External competitiveness is measured by the ratio of export prices to the prices charged by foreign competitors (measured in a common currency).
In theory, the most price competitive countries receive the most proportional change in demand. But it doesn't really make sense when demand falls. If competitiveness is a proportional thing, where the most competitive receives the highest proportion of demand, then that proportion might hold constant when demand falls - so they still receive the highest amount of a smaller demand number. Or... when demand falls, would a country ditch the most expensive product first, and leaving the cheapest products consuming a greater proportion of what turns out to be a smaller number.
... see the problem with ad hoc economic modelling, without theory?
-
Export Determination Models
These models explicitly recognise supply-side variables (e.g., export profitability, productive potential, etc.) may have an impact on the quantity exported; however, since they combine both the supply- and demand-side variables together into a single ad hoc equation, no allowance is made for any inter-relationship between the two sides of the market for determining the quantity of exports actually sold. A change in one of the supply variables will alter the quantity exported as if there were no demand constraint and vice versa.
-
System of Equations Models The third approach avoids the above-mentioned problems by specifying separate export demand and supply functions and estimating the resulting system of equations simultaneously. Exporting firms are all assumed to be operating in an identical environment at any given point in time, and the industry is then simply represented as the individual firm writ large.
The paper "Modelling Export Prices and Quantities in Selected Developing Economies.pdf" in the Documentation directory is an example of this.
In this Tuna World Trade Modelling work, all raw and tinned tuna are expressed in USD, and the "world price" is estimated. Each country's export prices are then expressed "relative to the world price" to get a sense of competitiveness. That way, changes in price relative to world price is also something that has economic meaning, as an expression of change in competitiveness (regardless of whether its domestic or exchange rate induced).
I'm not starting with a clean slate - the COMTRADE data showed lots of potential for a model, there are some changes I introduce:
-
Moving from a annual to a monthly/quarterly model
The previous modelling in the Comtrade project was annual. It was good, but I quickly started losing too many degrees of freedom when I tried to do some statistics. The trade data is monthly, so I should stick with the monthly values. This also raises the problem of the additional variables, and their time-dimensions, which I haven't a great answer for. I'm hoping something like IMF-based measures of GDP might help with 'size' measures and 'demand' measures.
-
Generate 'domestic price' and 'exchange rate' measures, such that the combination of the two make sense.
This is a bit fiddly, but there's value in being about describe international competitiveness using a common currency like the USD, and causes for changes in competitiveness like internal country price changes, and macro-economy induced changes coming through the exchange rate.
This will need some thinking.
-
Estimating a world price, at the USD level, made sense
Being able to express each country's competitiveness relative to a 'world price' is good.
-
Identifying Minnows over time, not at a specific time.
Through the COMTRADE work, there were large 'holes' in the models of raw and tinned tuna because of inconsistent timeseries at the country level. This was because the COMTRADE data identified minnows and non-minnows as countries who did or did not exceed 98% of the total volume.
This project will identify Minnows and non-minnows as companies which acheive 98% across the whole timeperiod.
-
Additional measures of country internal demand and export Supply
I'm thinking demand side metrics are GDP. And supplied side are ...? the fishing fleet size? Maybe, if we could have gotten fleet size for eall countries... Now I'm struggling and will need to think up something related to demand and supply which is still visible at the data level.
James Hogan, Senior Marine Resource Economist, FAME - SPC, 2026