Trends in Mexican Petroleum Production and Consumption

Mexico has recently been another top petroleum provider to the USA—number two according to Annual Energy Review for 2009 (see e.g., Fig. 5.4 and Table 5.4).  According to the EIA’s country analysis brief for Mexico, the oil sector still generates over 15 percent of the country’s export earnings, and, an incredible 40 percent of total government revenues rely on earnings from the oil industry.  

Recently though, Mexico’s production of oil has been decreasing substantially.  What if this trend continues?

The declines are in no small measure due to steep declines in Mexico’s Cantarell off-shore oil field.  Cantarell was once one of the largest producing oil fields in the world but in 2009, production was down 70 percent from it peak in 2004.  The decline in production from Cantarell, however, has been partially offset from increasing production in the “Ku-Maloob-Zaap” (KMZ) fields and some other offshore fields. 

According to the EIA country analysis brief for Mexico, the USA receives the “vast majority” of Mexico’s crude oil exports, mostly via tankers at the Gulf Coast.  The EIA has predicted that Mexico would become a net oil importer by 2015.  

Non-linear least squares (NLLS) analysis of total petroleum production

Figure 1 shows total petroleum production for 1965-2009 as reported in the BP statistical review.

The solid red line is the NLLS best fit to the full data set using the Hubbert equation with the parameters “a,” Qo and Q∞ allowed to vary to minimize the sum of the residual sums of squares (Srss).  The best fit (“a” = 0.09447; Qo = 2.02; Q∞ = 57.08) looks like it is skewed by the local peak which occurred in 1992.  Nevertheless, the down trend in production is quite clear, especially for the time span from 1999-2009, which is probably reflecting the peak in production (in about 2004-2005) and subsequent decline in the Cantarell field. 

Figure 1 also shows two other NLLS best fits to partial data sets using the Hubbert equation.  The pink line shows the best fit to the 1965-1986 time span (“a” = 0.2575; Qo = 0.09; Q∞ = 16.89).  The orange line shows the best fit to the 1999-2009 time span (“a” = 0.1773; Qo = 9.15; Q∞ = 31.21).  It is that later time span extrapolated to the future, which probably lead the EIA to predict that Mexico’s net exports will cease by about 2015.  More on this point later.

Modified analysis of the production data from 1980-2009

Similar to that done for the USA and Canada analysis in previous posts, I used the best fit values of “a”, Q_o and Q∞ obtained from fitting the 1965-86 data using Hubbert’s equation, shown in Figure 1, as “fixed” parameters for a subsequent fit to the 1987-2009 data, using equation [9], derived in Part 7 of the “Refining the Peak Oil Rosy Scenario” series, with “a”, Qo and Q∞ fixed to their best fit values from the 1965-1986 analysis and one or both fca and fcq varied.  The results are shown in Figure 2 below:

The solid blue line in Figure 2 shows the best fit to the 1965-1986 data using the Hubbert equation (once again, (“a” = 0.2575; Qo = 0.09; Q∞ = 16.89).  The solid green line shows the best fit obtained to the 1987-2009 data using equation [9] with fca = 0.0928 and fcq = 1.09. 

The overall Srss obtained using the Hubbert equation to the full data set (1965-2009) equaled 0.7925.  The sum of the Srss for the Hubbert equation fit to 1965-86 (Srss = 0.1056) plus the modified equation [9] best fit to 1987-2009 (Srss = 0.5636) with only fcq varied (best fit fcq=1.042) was significant less (F-test, p<0.01). 

However, the sum of the Srss for the Hubbert equation fit to 1965-86 (Srss = 0.1056) plus the modified equation [9] best fit to 1987-2009 (Srss = 0.0726) with both fcq and fca varied gave a still better fit than the fit with only fcq varied (F-test, p<0.0001).

These best fit parameters from fca and fcq imply that for the 1987-2009 time span there is a trend for the total extractable oil (Q∞) to increase 9 percent per year while the rate constant for production (“a”) is decreasing by 7.2 percent per year.  The increase in Q∞ does not overcome the decrease in “a” and therefore there is an overall trend for a decrease in the production rate (dQ/dt) through the 1987-2009 time span.  This trend is interrupted by the spike in production centered at 2004, however.

The trend based on the modified equation [9] analysis of the 1987-2009 data predicts the production rate decreasing from a peak in 2003, which is about the same as the actual peak in 2004.  It also predicts the projected decline curve being shallower than that predicted from the Hubbert equation best fit of the full 1965-2009 time span or to the actual 1999-2009 data.  The modified fit using equation [9] does not do too good a job of modeling those most recent years of production data, which first over shoots and then undershoots the equation [9] prediction curve.   

Therefore in my subsequent analysis, I have focused more on the results of the Hubbert fit to the 1999-2009 time span as this may be a better predictor of the near-term future production trends for Mexico. 

Non-linear least squares (NLLS) analysis of total petroleum consumption
Figure 3 shows the best fit to Mexican petroleum consumption data (as reported in the BP statistical review):

The red line in Figure 3 shows the best-fit of the Hubbert equation to the full data set 1965-2009 from the BP statistical review (“a” = 0.0755; Qo =1.74;  Q∞ = 37.89). 

The fit to the full time span of data looks much better than the comparable full data fits to USA or Canadian consumption data, most likely because the spikes in consumption around 1980 and 1990 are very mild or non-existent in Mexico.   

The best fit to the consumption data suggests that the rate of consumption peaked in 2005,  close to the same year as production peaked, and is now declining. 

I find it interesting that the consumption rate constant (“a”) for Mexico equal to about 7.5%/yr.  This is nearly double that of the USA’s (4.0% /yr for the 1984-2009 time span) and Canada’s (4.4% year for the 1991-2009 time span).  Perhaps large rate of increase in petroleum consumption reflected rapid modernization (and ensuing increased use of petroleum products) of Mexico, as afforded by its wealth created from exporting oil.  For instance, back in 1965, Mexico’s production was 30% of Canada’s and less than 3% of USA’s production, whereas at its peak in 2004, Mexico’s production rate was about 60% of Canada’s and 30% of the USA’s total production.  That was an enormous relative increase in “petroleum wealth” compared to the USA and Canada over a 4 decade period.  My impression is that, sadly, Mexico doesn’t have much to show for this, and, the situation is about to change for the worse.

Comparing future trends in Mexican petroleum production and consumption

Figure 4 shows the best fits obtained using the Hubbert equation analysis of the total time span of Mexican consumption data and the 1999-2009 production data, and, my modified analysis of the 1987-2009 production data (dash line) using equation [9]. 

The predicted export curve is shown based on both the prediction based on 1999-2009 production data trend (solid green) and based on the prediction based on the 1987-2009 production data trend (dash green).  Additionally, I show the actual “measured” export data from 1987-2009 (i.e., the BP statistical review reported production minus reported consumption).

The extrapolated export trend predicts that if the Mexican production trend from 1999-2009 continues, then Mexico will have no net exports by 2015.  This is consistent with the EIA’s prediction that EIA Mexico would become a net oil importer by 2015. 

On the other hand, if the Mexican production trend from 1987-2009 proves to be a more accurate predictor then, the decline in exports is predicted to be much more gradual still providing net exports but diminishing exports to 2030 and beyond.  Even if the predicted trend based on the 1987-2009 production data turns out to be more accurate in the long run, Mexico’s exportable production in 2015 would still be down by about 20% compared to 2009 and down about 35% in 2020 compared to 2009. 

My expectation is that in the long run, exports will be somewhere in-between what is predicted from the 1999-2009 production trend and what is predicted by the 1987-2009 production trend.  That is the production trend will reverts back towards the longer 1987-2009 trend, e.g., because the rate of decline in Cantarell will become less steep, or, there is increasing production in the KMZ or other offshore fields, or both. 

BUT, even if the steep decline predicted from the 1999-2009 data analysis continues only out to 2015 and then reverted to a shallower decline trend predicted from the 1987-2009 data analysis, it won’t matter because the amount of petroleum produced at that point would already be below the domestic consumption rate, and therefore, provide no net petroleum exports.

Only time will tell which of these trends prevails, but what seems undeniable to me is that the years when Mexico could export plus 0.5 bbs/yr (or about 1.3 mb/d), mostly to the USA, are over. 

--Updated Nov 28, 2010: as explained in Canada—Petroleum Superpower or Super-slave? --

Impact on USA

Figure 5 reproduces the USA production and consumption trends, plus Canadian exportable petroleum from the previous article, “Trends in Canadian petroleum production and consumption,” and add to it, the predicted Mexican exportable petroleum assuming that the production trend from 1999-2009 continues (solid green line).   For reference, I also include the measured export data for 1991-2009 for Mexico (green circles) and similar measured exportable petroleum data for Canada (open brown circles).

The solid green line then depicts the sum of predicted USA production and exports from Canada and Mexico.  This prediction trend and the past measured export data, illustrate that, although it has been a major exporter of oil to the USA for the past decade, Mexico’s oil exports (based on the decline trend predicted from the best fit to the 1999-2009 production) are about to end.  Going forward, the oil predicted to be exported to the USA from Mexico will be a pittance of what Mexico used to export to the USA, or, what the USA is predicted to need.

For 2010, I predict that Canadian and Mexican exports together with USA’s domestic production will provide about 48 percent of the USA’s predicted consumption.  But by 2015, with no Mexican exports predicted, we are back to scenario where USA’s production plus Canadian exports is predicted to provide 42 percent of the USA’s consumption needs, with nothing from Mexico.

In the coming years following 2015 I predict that Mexico will provide little to nothing to meet the USA's petroleum consumption.

Implications for Mexico

Although this article is mainly focused on the petroleum import implications for the USA, I think it is worthwhile to briefly consider what impact the cessation of being an oil exporting county, from 2015 on, will have on Mexico.  Because Mexico is the USA next-door neighbor, what happens in Mexico will affect the USA.

Perhaps becoming a net importer will not be too much of a shock for the people of Mexico, as it is already a net importer of refined petroleum products such as gasoline.  For instance, as reported in the EIA country report, in 2009 Mexico imported 519,000 bbl/d of refined petroleum products, while exporting 244,000 bbl/d, and, gasoline represented about 60 percent of products imported.  

But going forward past 2015, what will pay for those refined petroleum products?  Certainly not money earned from the net export of petroleum, as there will likely be little to no net exports.  If some of the petroleum produced has to be sold for the import of refined products, then this will cut into the amount of petroleum available for domestic consumption.

As I noted above, the EIA estimated that 40 percent of total government revenues rely on earnings from the oil industry.  If there are no net exports, then I expect that this will strongly negatively impact government income.  The Mexican government’s ability to provide basic service and maintain the country’s infrastructure will accordingly severely tested.   Should there be a further breakdown in these services and infrastructure in Mexico, this will surely have a negative impact on the USA.

Trends in Canadian Petroleum Production and Consumption

Canada has consistently been the top petroleum provider to the USA.  According to the EIA’s country analysis brief for Canada over 99 percent of the oil that Canada exports, it exports to the USA mainly through an extensive pipeline system interconnecting the two countries. 

Non-linear least squares (NLLS) analysis of total production

Figure 1 shows total petroleum production for 1965-2009 as reported in the BP statistical review: 

The solid red line is the NLLS best fit to the full data set using the Hubbert equation with the parameters “a,” Qo and Q∞ allowed to vary to minimize the sum of the residual sums of squares (Srss).  Because there is not enough sign of the production data reaching a plateau, the Hubbert equation “blows up” and gives an unrealistically high estimate for Q∞ (“a” = 0.0228; Qo = 19.27; Q∞ = 30015). 

The solid pink line is the NLLS best fit to the full data set with Q∞ fixed equal to 178, the estimated proven reserves for Canada, as reported in the EIA country analysis brief.  “Proven reserves,” however, have sometimes been notoriously underestimated (because trends for improvements in extraction technology may not be considered) or overestimated (to overcome OPEC production quotas in the Middle East), so I am not too confident in such numbers.    

The total production data for Canada showed a local peak in 1973, similar to the USA total production which peaked in 1970.  However, whereas the Canadian production started to increase again after a minimum in 1976, the USA total production went into a steady decline.  Overall, the trend in Canadian production is most like the trend seen in NGL production in the USA which had a local peak in 1972, a minimum in the late 70s and steady increase thereafter. 

I adopted the same approach that I used to analyze the NGL production in the USA. 

Modified analysis of the production data from 1980-2009

I used the best fit values of “a”, Q_o and Q∞ obtained from fitting the 1965-76 data using Hubbert’s equation also shown in Figure 1, as “fixed” parameters for a subsequent fit to the 1978-2009 data, using equation [9], derived in Part 7 of the “Refining the Peak Oil Rosy Scenario” series, with “a”, Qo and Q∞ fixed to their best fit values from the 1965-1976 analysis and one or both fca and fcq varied.  The results are shown in Figure 2 below:

The solid blue line in Figure 2 shows the best fit to the 1965-1976 data using the Hubbert equation (“a” = 0.0966; Qo = 1.26; Q∞ = 11.69).  The solid green line shows the best fit obtained to the 1977-2009 data using equation [9] with fca = 0.0968 and fcq = 1.052. 

The overall Srss obtained using the Hubbert equation to the full data set (1965-2009) equaled 0.2069.  The sum of the Srss for the Hubbert equation fit to 1965-77 (Srss = 0.0233) plus the modified equation [9] best fit to 1978-2009 (Srss = 0.0932) with only fcq varied (best fit fcq=1.039) was significant less (F-test, p<0.0001). 

However, the sum of the Srss for the Hubbert equation fit to 1965-77 (Srss = 0.0233) plus the modified equation [9] best fit to 1978-2009 (Srss = 0.0294) with both fcq and fca varied gave a still better fit than the fit with only fcq varied (F-test, p<0.0001).

These best fit parameters from fca and fcq imply that for the 1978-2009 there is a trend for the total extractable oil (Q∞) to increase 5.2 percent per year while the rate constant for production (“a”) is decreasing by 3.2 percent per year.  The increase in Q∞ overcomes the decrease in “a” to give an overall increase in the production rate (dQ/dt) through the 1978-2009 time span. 

The trend based on the modified equation [9] analysis predicts that the production rate will continue to increase over the next 20 years.  In fact, the projection based on the modified analysis is nearly the same as that predicted from the Hubbert equation best fit that “blew up” (dashed red line in Figure 2) up until about 2020 but then shows signs of a declining rate of increase in production thereafter.  The predicted production rate is substantially higher than what one might have expected assuming that in Q∞ equals 178 (dashed pink line in Figure 2).
 

Non-linear least squares (NLLS) analysis of total consumption

Despite being a net exporter, Canada is still a significant importer of petroleum (about 1.2 bbl/d of crude oil and refined products in 2008) apparently because its major eastern population centers are not connected to production facilities in the west.  The EIA country analysis brief states that most of Canada’s oil imports come from Algeria and Norway (crude oil) and the U.S. (refined products).

Figure 3 shows the best fit to Canadian petroleum consumption data (as reported in the BP statistical review). 

The short-dash red line in Figure 3 shows the best-fit of the Hubbert equation to the full data set 1965-2009 from the BP statistical review (“a” = 0.0206; Qo =28.60;  Q∞ = 173).  Similar to the USA consumption data, local peaks in consumption in 1979 and again in 1990 tend to skew the best fit to a shallower curve than suggested by the more recent data. 

The long dashed red line in Figure 3 shows the best fit of the Hubbert equation to the more limited data set 1984-2009, which shows the more recent trend in consumption after the peak in consumption in 1979 and decline through 1983.  Unfortunately, this best fit “blew up” resulting in an unrealistically high estimate of Q∞ (e.g., Q∞ = 5195) and low estimate of “a” (“a”=0.0167). 

Finally, the solid red line shows the best fit to the last 18 years of data from 1991-2009 (“a” = 0.0445; Qo =16.54;  Q∞ = 81.9), which I have take to be the most reasonable estimate of Canadian consumption of the three fits depicted in the figure.   The predicted consumption curve from 2010 to 2030 is not all that different from that predicted from the best fit to the full data set.  Although, the consumption max in 2021 of 0.91 bbs/yr would be years early than that predicted from the full data set. 

It is interesting to note that the total predicted maximum petroleum consumed of about 82 bbs is about 11% percent of the total predicted maximum petroleum consumed (about 744 bbs) predicted for the USA.  This is roughly proportional to the relative populations of the USA (307 million) and Canada (33 million), suggesting that over the long run, Canada’s and the USA’s utilization of petroleum would be about the same, if these consumption trends were to continue. 

--Updated Nov 28, 2010: as explained in Canada—Petroleum Superpower or Super-slave? --

Comparing future trends in Canadian petroleum production and consumption

The extrapolated trend predicts Canada to continue to produce more than it consumes for at least the next 20 years.  The net difference between the predicted production and consumption is shown as the dashed brown line and open symbols. 

As discussed in the article Canada—Petroleum Superpower or Super-slave?  The total net exports to the USA equals the 73 percent of total petroleum production send from Canada to the USA minus 12 percent of total petroleum consumption sent to Canada from the USA. 

Net exports to the USA are predicted to increase about 42 percent from the present amount of about 0.8 bbs/yr to 1.14 bbs/yr in 2030.  This is mainly because of the predicted increase in production as well as the plateau in domestic consumption.  In about 2013 the net amount of petroleum exported to the USA is predicted to actually exceed domestic Canadian consumption.

Impact on USA

This is good news for the USA, since as indicted by the EIA country analysis brief, most of Canada’s petroleum production goes to the USA, and as discussed in Canada—Petroleum Superpower or Super-slave?  The proportion to the available to USA can be decreased in accordance with the NAFTA proportionality clause.

Moreover, I would expect that this trend would continue going forward because the cost to export oil anywhere else would be more expensive than just pumping it through pipelines to the USA. 

But just how much does of this bounty of Canadian oil mitigate the downward trend in USA production and the great disparity between USA production and consumption? 

To answer this question, I prepared Figure 5, which reproduces the USA production and consumption trends shown in Figure 11 from the previous article, “Trends in USA petroleum production and consumption.” I further modified this graph to plot the predicted USA production plus Canadian exported petroleum from Figure 4 above.

 

The brown line, showing USA production plus Canadian exportable petroleum, illustrates that the Canadian petroleum alone will not do much to serve to predicted USA consumption for 2010-2030.  The Canadian exports do tend to mitigate and then flatten out the decline in USA production from 2020-2030 but still, the sum of petroleum provided is far below what the USA is projected to consume.  For instance, is the plateau region from 2020-2030 USA production plus Canadian exportable petroleum is predicted to provide the USA with about the same amount of petroleum that the USA consumed in 1955.  But in 1955 the USA’s population was 165 million http://www.census.gov/popest/archives/1980s/st5060ts.txt or about 54 percent of the present 307 million population. 

Can the USA’s other next door neighbor, Mexico, makeup for some of the shortfall to the USA? 

This is explored in the next post....